FreshRSS

🔒
❌ About FreshRSS
There are new available articles, click to refresh the page.
Before yesterdayYour RSS feeds

Alphabet X’s exosuit

By Brian Heater

Last week, Kathryn Zealand shared some insight on the eve of Women’s Equality Day. The post highlighted an issue that’s been apparent to everyone in and around the robotics industry: there’s a massive gender gap. It’s something we try to be mindful of, particularly when programming events like TC Sessions: Robotics. Zealand cites some pretty staggering figures in the piece.

According to the stats, around 9% of robotics engineers are female. That’s bad. That’s, like, bad even by the standards of STEM fields in general — which is to say, it’s really, really bad. (The ethnic disparities in the same source are worth drawing attention to, as well.)

Zealand’s piece was published on LinkedIn — fitting, given that the overarching focus here is on hiring. Well worth your time, if you’re involved in the hiring process at a robotics firm and are concerned about broader diversity issues (which hopefully go hand in hand for most orgs). Zealand offers some outside of the box thinking in terms of what, precisely, it means to be a roboticist, writing:

We have a huge opportunity here! Women and other under-represented groups are untapped pools of talented people who, despite not thinking of themselves as “roboticists,” could be vital members of a world-changing robotics team.

I’m going to be real with you for a minute, and note what really caught my eye was that above image. See, Zealand is a Project Lead at Alphabet X. And what you have there is a robotic brace — or, rather, what appears to be a component of a soft exosuit.

Image Credits: Bryce Durbin/TechCrunch

Exosuits/exoskeletons are a booming category for robotics right now that really run the gamut from Sarcos’ giant James Cameron-esque suit to far subtler, fabric-based systems. Some key names in the space include Ekso Bionics, ReWalk and SuitX. Heck, even Samsung has shown off a solution as part of a robotics department that appears to be largely ornamental at the moment.

Image Credits: Harvard Biodesign Lab

Most of these systems aim to tackle one of two issues: 1) Augmenting workers to assist with difficult or repetitive tasks for work and 2) Provide assistance to those with impaired mobility. Many companies have offers for both. Here’s what Harvard’s Biodesign Lab has to say on the matter:

As compared to a traditional exoskeleton, these systems have several advantages: the wearer’s joints are unconstrained by external rigid structures, and the worn part of the suit is extremely light. These properties minimize the suit’s unintentional interference with the body’s natural biomechanics and allow for more synergistic interaction with the wearer.

Alphabet loves to give the occasional behind-the-scenes peak at some of its X projects, and it turns out we’ve had a couple of glimpses of the Smarty Pants project. Zealand and Smarty Pants make a cameo in a Wired UK piece that ran early last year about the 10th anniversary of Google/Alphabet X. The piece notes that that the project was inspired by her experience with her 92-year-old grandmother’s mobility issues.

Image Credits: Alphabet X

The piece highlights a very early Raspberry Pi-controlled setup created by a team that includes costume designers and deep learning specialists (getting back to that earlier discussion about outside the box thinking when it comes to what constitutes a roboticist). The system is using sensors in an attempt to effectively predict movement in order to anticipate where force needs to be applied for tasks like walking up stairs. The piece ends on a fittingly somber note, “Fewer than half of X’s investigations become Projects. By the time this story is published it will probably have been killed.”

My suspicion is that the team is looking to differentiate itself from other exosuit projects by leveraging Google’s knowledge base of deep learning and AI to build out those predictive algorithms.

Alphabet declined to offer additional information on the project, noting that it likes to give its moonshot teams, “time to learn and iterate out of the spotlight.” But last October, we got what is probably our best look at Smarty Pants, in the form of a video highlighting Design Kitchen, Alphabet X’s lab/design studio.

Image Credits: Alphabet X

The Wired piece mentions a “pearlescent bumbag,” holding the aforementioned Raspberry Pi and additional components. For you yanks, that’s a fanny pack, which are not referred to as such in the U.K., owing to certain regional slang. Said fanny pack also makes an appearance in the video, providing, honestly, a very clever solution to the issue of hanging wires for an early-stage wearable prototype.

“One of the things that’s really helped the team is being really focused on a problem. Even if you spent months on something, if it’s not actually going to achieve that goal, then sometimes you honor the work that’s been done and say, ‘we’ve learned a ton of things during the process, but this is not the one that’s actually going to solve that problem.’ ”

The most notable takeaway from the video is some additional footage of prototypes. One imagines that, by the time Alphabet feels confident sharing that sort of stuff with the world, the team has moved well beyond it. “It doesn’t matter how janky and cardboard-and-duct-tape it is, as long as it helps you learn — and everyone can prototype, even while working from home,” the X team writes in an associated blog post.

The one other bit of information we have at the moment is a granted patent application from last year, which comes with all of the standard patent warnings. Seeing a patent come to fruition is often even more of a longshot (read: moonshot) than betting on an Alphabet X project to graduate. But they can offer some insight into where a team is headed — or at least some of the avenues it has considered.

Image Credits: Alphabet X

The patent highlights similar attempts to anticipate movement as those highlighted above. It effectively uses sensors and machine learning to adjust the tension on regions of the garments designed to assist the wearer.

Image Credits: Alphabet X

The proposed methods and systems provide adaptive support and assistance to users by performing intelligent dynamic adjustment of tension and stiffness in specific areas of fabric or by applying forces to non-stretch elements within a garment that is comfortable enough to be suitable for frequent, everyday usage. The methods include detecting movement of a particular part of a user’s body enclosed within the garment, determining an activity classification for that movement, identifying a support configuration for the garment tailored to the activity classification, and dynamically adjusting a tension and/or a stiffness of one or more controllable regions of the garment or applying force to non-stretch fabric elements in the garment to provide customized support and assistance for the user and the activity the user is performing.

It’s nice seeing Alphabet take a more organic approach to developing robotics startups in-house, rather than the acquisitions and consolidations that occurred several years back that ultimately found Boston Dynamics briefly living under the Google umbrella. Of course, we saw the recent graduation of the Wendy Tan White-led Intrinsic, which builds software for industrial robotics.

All right, so there’s a whole bunch of words about a project we know next to nothing about! Gotta love the startup space, where we’re definitely not spinning wild speculation based on a thin trail of breadcrumbs.

I will say for sure that I definitely know more about Agility Robotics than I did this time last week, after speaking with the Oregon-based company’s CEO and CTO. The conversation was ostensibly about a new video the team released showcasing Digit doing some menial tasks in a warehouse/fulfillment setting.

Some key things I learned:

  1. Agility sold a dozen Cassie robots, largely to researchers.
  2. It’s already sold “substantially more” Digits.
  3. The team includes 56 people, primarily in Oregon (makes sense, as an OSU spinout), with plans to expand operations into Pittsburgh, everyone’s favorite rustbelt robotics hub.
  4. Agility is consulting with “major logistics companies.”
  5. In addition to the Ford delivery deal, the company has its sights set on warehouse tasks in hopes of offering a more adaptable solution than ground-up warehouse automation companies like Berkshire Gray.

Image Credits: Agility Robotics

Oh, and a good quote about job loss from CEO Damion Shelton:

The conversation around automation has shifted a bit. It’s viewed as an enabling technology to allow you to keep the workforce that you have. There are a lot of conversations around the risks of automation and job loss, but the job loss is actually occurring now, in advance of the automated solutions.

Agility hopes to start rolling out its robots to locations in the next year. More immediate than that, however, is this deal between Simbe Robotics and midwestern grocery chain, Schnuks. The food giant will be bringing Simbe’s inventor robots to all of its 111 stores, four years after it began piloting the tech.

Schnuck Markets deploys Tally robot by Simbe Robotics to its stores – bringing shelf insights for better shopping experience. Photographed on Friday, Aug. 13, 2021, in Des Peres, Mo.

Simbe says its Tally robot can reduce out of stock items by 20-30% and detect 14x more missing inventor than standard human scanning.

Carbon Robotics (not to be confused with the prosthetic company of the same name that made it onto our Hardware Battlefield a few years back) just raised $27 million. The Series B brings its total funding to around $36 million. The Seattle-based firm builds autonomous robots that zap weeds with lasers. We highlighted their most recent robot in this column back in April.

And seeing how we recently updated you on iRobot’s continued indefinite delay for the Terra, here’s a new robotic mower from Segway-Ninebot.

Image Credits: Segway-Ninebot

Segway’s first robotic lawnmower is designed for a lawn area of up to 3,000 square meters, has several features of a smart helper in the garden and is the quietest mower on the market with only 54 dB. The Frequent Soft Cut System (FSCS) ensures that the lawn is cut from above and the desired height is reached gradually. Offset blades allow cutting as close as possible to edges and corners.

That’s it for the week. Don’t forget to sign up to get the upcoming free newsletter version of Actuator delivered to your inbox.

AxleHire to scale Tortoise and URB-E zero-emissions delivery solutions nationally

By Rebecca Bellan

Last-mile logistics supplier AxleHire provides same-day and next-day delivery through a network that includes gig economy, couriers and traditional carriers. Over the past year, it has been quietly piloting automated repositioning startup Tortoise’s remote controlled delivery robots in Los Angeles and compact container delivery service URB-E’s e-bike container delivery in New York City. On Thursday, it announced plans to scale the two very different zero-emissions pilot programs nationally over the next 12 months.

AxleHire, which is known for parcel delivery and restaurant meal kit delivery like Blue Apron and HelloFresh, plans to bring over 100 Tortoise robots across the country. During URB-E’s summer deployment with AxleHire in NYC, it deployed 10 vehicles moving 100 containers per week. Now it will deploy 50 URB-E vehicles moving anywhere from 300 to 500 containers per week in NYC, LA and San Francisco, as well as other launch cities. The company, which raised a $20 million round in April, didn’t specify every city it would be entering with these new programs, but Tortoise and URB-E said we can look to the cities AxleHire already operates in: Chicago, Dallas, Houston, Los Angeles, San Diego, San Francisco, New York, Phoenix, Seattle and Portland, Oregon.

AxleHire’s style is to establish delivery hubs in or near dense metro areas, which makes for easier trips and less miles traveled in total. The partnerships with Tortoise and URB-E are a part of AxleHire’s mission to create more sustainable and cheaper last-mile delivery. The company says its partnerships with the two startups have also lowered its emissions by 95%. AxleHire is providing an example of one company trialing two very different greener and tech-focused forms of transporting goods, so it will likely serve as an interesting case study for other last-mile logistics providers.

Image Credits: URB-E

In New York, AxleHire and URB-E have been working together on a microcontainer delivery system between Brooklyn and Manhattan. URB-E’s vehicles are specifically designed to be able to ride in the bike lanes, despite their ability to haul over 800 pounds. AxleHire says its pilot with URB-E resulted in a six times reduction in traffic and a model that is three times cheaper than EV delivery vans, largely based on the avoidance of parking tickets.

Over the past year in Los Angeles, AxleHire stationed Tortoise’s electric, 4-mph remote-piloted carts, which carried up to 120 pounds worth of goods, in its delivery microhubs in cities, allowing the little bots with friendly smiley faces to go back and forth, making about 15 deliveries per day within a three-mile radius. In addition, AxleHire loaded a large truck with multiple packages and a Tortoise robot, which would then drive into a dense residential area. This truck would serve as a mobile delivery hub, doing its own deliveries while the bot goes back and forth delivering parcels and being reloaded all day long.

“It’s basically the hive model, where we’re augmenting the existing van or truck in terms of how many deliveries they could do in a two-hour stretch,” Dmitry Shevelenko, co-founder of Tortoise, told TechCrunch. “There’s communication happening with our subject confirming they’ll be home to receive it. If so, they get notified that the robot’s on the way when it’s about 10 minutes away, and then when it arrives, the customer will come out and get it from the containers in the robot.”

The Tortoise bots, which can ride on sidewalks or bike lanes, have both swappable batteries and can be plugged and charged, according to Shevelenko. On a single charge, they can get around 10 to 15 miles of range.

While Tortoise’s bots will be operated 100% remotely over the next year, remote positioning is not Tortoise’s end goal at all. Autonomy is the goal, and doing partnerships like this, as well as with shared e-scooter operators like Spin, allows Tortoise not only to get into markets that currently don’t have regulation for self-driving vehicles, but also to just get into the market now, rather than spending multiple years mapping it first. The only real infrastructure the bots need is 4G connectivity.

“The beauty is that we can ship the robot to a new location and because we have the benefit of human judgment oversight every inch of the journey,” said Shevelenko. “We don’t need perfect routing or perfect mapping. We’re filling in the maps over time, and that gives us a big data advantage.”

By slowly collecting routing data over the course of the next year, Tortoise will be giving its system more data to learn on and create the most optimal route for the specific use case of low-speed and lightweight delivery vehicles. Shevelenko says the long-term vision of Tortoise is to have its tech on any light electric vehicle, whether it be a delivery robot, a scooter, a cleaning robot, security robot or construction robot. Delivery is a great place to start, given the massive demand in the COVID marketplace.

“The more vehicles we have with Tortoise eyes on them, the more data we’re collecting, which means we’re doing trips with higher autonomy and lower costs,” said Shevelenko.

Aside from allowing for max data collection, remote controlled delivery bots over the next year also give Tortoise the advantage of getting the community used to this new tech.

“We think the right way to enter a community is first to reassure people that this is safe and get them comfortable with it,” said Shevelenko. “Once it’s part of daily life, then slowly over time, we can turn on more autonomy, but there’s no need to rush into that right now. The practical reality is, everybody’s claiming they’re doing autonomy but they aren’t. They always have a fallback like safety drivers or remote monitors. Nobody actually trusts their economy system, and so we’re kind of leaning into that and not trying to do something that is impossible.”

Agility Robotics’ Digit gets a warehouse gig

By Brian Heater

A new video from Agility Robotics showcases an increasing familiar sight: advanced, autonomous robots performing boring warehouse tasks. It’s not the sort of video that tends to be hugely viral for a company, rather, it’s the sort of meat and potatoes proof of concept that companies like Boston Dynamics wedge between flashy videos of parkour and highly choreographed dance sessions.

Ultimately, however, this is precisely the sort of tasks the robots’ creators are targeting: the well-known trio of dull, dirty and dangerous. Moving payloads back and forth certainly ticks that first box pretty well. There’s a reason warehouse and fulfillment workers often liken their work to robotics.

“The conversation around automation has shifted a bit,” Agility CEO Damion Shelton tells TechCrunch. “It’s viewed as an enabling technology to allow you to keep the workforce that you have. There are a lot of conversations around the risks of automation and job loss, but the job loss is actually occurring now, in advance of the automated solutions.”

Digit, the bipedal robot the company announced back in 2020, had its most high-profile moment in the spotlight after Agility announced a partnership with auto giant Ford back at CES. The auto giant currently owns two of the robots, with long-term plans to utilize the technology for delivery.

Today’s video is an attempt to showcase some more short-term solutions, putting Digit to work on more menial tasks.

Image Credits: Agility Robotics

“The value and goal of a machine like Digit is the generality,” CTO Jonathan Hurst says. “It’s a robot that can operate in human environments and spaces. It’s a relatively straightforward thing for very structured, repetitive tasks, to say, ‘there’s going to be boxes over there. We’re going to tell you which one from a databasing system, and we want you to move it over there.’ Maybe this is something that it does for three or four hours a day and then it goes to a different space and does it three or four hours and then it unloads a tractor trailer.”

The company sees Digit’s value as a more plug and play solution than something like Berkshire Gray’s offerings, which builds a fully automated warehouse from the ground up. There’s still programming involved, of course. An Agility rep will appear on-site to pre-map a location and help the robot execute its repetitive tasks.

“In terms of where we can actually deploy and do useful work for a customer, it turns out a lot of the tasks — walk from Point A to Point B, pick up and carry a package — are portable across these environments,” says Shelton. “There’s no real core piece of technology that you develop, that’s different for an indoor environment versus outdoor. It’s just the level of maturity. I think we’ve reached that pretty quickly on the indoor stuff, so it’s a logical first place for deployment.”

Image Credits: Agility Robotics

Agility hasn’t announced partners beyond Ford, though it says it’s currently working with “major logistics companies.” It hasn’t revealed numbers of Digits sold, either, though it tells TechCrunch that the number is “substantially more” than the dozen Cassie units it sold prior to Digit, largely for research purposes. Sales are largely CapEx at the moment, though the company is exploring other opportunities, such as a RaaS (robotics-as-a-service model).

Agility’s team is currently at 56 people, primarily based in Oregon (the company got its start as part of OSU’s nascent robotics division), where the robots are primarily manufactured.

“We’ve grown pretty rapidly since last December,” says Shelton. “We’re expanding our Pittsburgh office by the end of the year, in addition to the Oregon office. We have a pretty rapid growth rate. As we’ve been increasing the production rate on the robots, we’ve had a fair amount of hiring for that. We just moved into a new facility that we remodeled, back in June.”

 

Carbon Robotics secures $27M for its autonomous field weeders

By Brian Heater

Agricultural robotics firm Carbon Robotics (not to be confused with our former Battlefield contestant) announced this week that it has secured $27 million in funding. The round — which features Anthos Capital, Ignition Capital, Fuse Venture Partners and Voyager Capital — follows an $8.4 million Series A raised back in 2019. The company’s total funding is now at around $36 million.

“Weeding is one of the biggest challenges farmers face, especially with the rise of herbicide-resistant weeds and increasing interest in organic and regenerative methods,” founder and CEO Paul Mikesell said in a release. “This round of investment will enable us to scale our operations to meet the increasing demand for this technology. Additionally, this funding will allow our team to continue to innovate new products and identify revolutionary ways to apply technology to agriculture.”

The Seattle-based startup’s primary offering is an autonomous robot that uses lasers to zap weeds. The round follows the April announcement of Carbon’s latest-generation Autonomous Weeder, which it says is capable of eradicating around 100,000 weeds per hour. The pandemic has continued to accelerate interest in many agricultural robotics companies, as labor shortages continue to mount.

Carbon notes some international bans on various pesticides have left many farmers searching for an alternative solution. A system that works without the need for harmful chemicals that also reduces human labor in an industry often suffering from shortages in headcount has clear appeal.

The company says it has already sold out of its 2021 and 2022 stock, so one assumes scaling up production and headcount will be key investments from this round.

Simbe’s robots will be deployed across midwestern grocery chain, Schnucks

By Brian Heater

St. Louis-based grocery chain Schnucks (one of those “With a name like Smucker’s, it has to be good” situations, one imagines) announced this week that it will be deploying technology from Simbe Robotics across its 111 U.S. locations.

The deal comes a year and a half into a global pandemic that has substantially increased interest in automation, particularly around essential businesses – a qualifier that certainly applies to grocery stores.

Simbe’s mobile robots provide inventory scanning, offering a constantly updating picture of what’s on the store shelves and what needs to be restocked. Anyone who’s ever worked retail can almost certainly tell you that doing inventory is one of the biggest headaches in the industry, often requiring hours-long shutdowns or overnight marathons to complete.

The “multi-year” chain-wide rollout comes four years after Schnucks first began piloting the tech. Over the years, the partnership has gradually expanded. Simbe says its shelf-scanning Tally robot is capable of reducing out of stock items by 20-30% and detect 14 times more missing inventory than traditional human scanning.

Schnuck Markets deploys Tally robot by Simbe Robotics to its stores – bringing shelf insights for better shopping experience. Photographed on Friday, Aug. 13, 2021, in Des Peres, Mo.

“By deploying Tally to all stores, we are fully operationalizing these insights into our supply chain and expanding our ability to leverage real-time data to make revenue impacting decisions,” Schnucks VP Dave Steck said in a release. “Tally has become an integral component of our stores, streamlining operations and ultimately creating a better store experience for our customers and teammates.”

A number of companies are working to automate the world of inventory scanning, including Brain Corp and Bossa Nova, though the latter was dealt a massive setback when Walmart ended a large contract at the end of 2020.

I don’t know what to do with those tossed salads and robot legs

By Brian Heater

We’re becoming a newsletter soon! Please sign up to get Actuator in your inbox as soon as the first issue hit!

One of the most fascinating aspects of Boston Dynamics’ transition into a commercial organization is watching the company — and its partners — figure out real-world jobs for Spot. There’s no question that the tech is impressive, but there’s always been the broader subject of usefulness beyond the company’s initial purpose of serving as off-road pack mules.

We’ve seen some interesting examples since Spot first went on sale, including inspection for constructions sites and potentially dangerous settings — from nuclear power plants to off-shore oil rigs. There have also been some, shall we say, more controversial gigs, including Spot’s time as an electronic K9 for the NYPD.

But maybe finding the perfect job for Spot entails thinking both outside the box and Earth’s gravitational pull. NASA’s JPL in California has been working with the quadrupedal robot for a couple of years now, first as part of a DARPA challenge and now as a potential way to explore extraterrestrial caves. For this week’s installment of Actuator, we spoke to JPL NeBula Autonomy Project lead Ali Agha about the partnership.

How long has NASA been working with Spot?

We have been working with the SPOT robots for about two years now. We initially integrated our NeBula autonomy and AI solutions on the Spot robot as one of our robots participating in the DARPA Subterranean challenge competition. However, since then we have extended the application of these robots and JPL’s NeBula autonomy solution to planetary cave exploration and surface exploration as well as terrestrial disaster response and mining efforts.

What is the advantage of using legs (as opposed to wheels) on the Martian surface?

Imagine a no-road terrain on Earth. The ability to walk will allow traversing different elements of such a terrain much better than a typical wheeled vehicle. Similarly, legged locomotion can potentially enable totally new missions when exploring extreme and challenging terrains on planetary bodies in the solar system beyond our home planet.

How closely does NASA/JPL work with a company like Boston Dynamics on a project like this?

We have had an amazing collaboration with Boston Dynamics and work closely with them. On our project, JPL and Boston Dynamics’ efforts are highly synergistic. At JPL, we develop autonomy and AI solutions (called NeBula) acting as the robot brain to enable fully autonomous exploration of extreme and challenging environments with very minimal (to none) prior information about the terrain or environmental conditions.

NeBula is agnostic to the choice of robotic platform and can be used on wheeled rovers, legged platforms, as well as drones. On the other hand, Boston Dynamics is developing cutting-edge incredible robotic locomotion systems that can maintain the stability of the system over extreme environments. As a result, the combination of an autonomy solution like NeBula with a capable locomotion system like Boston Dynamics’ Spot opens up avenues for totally new classes of planetary and terrestrial missions.

I know autonomy is a big piece of this. Do the robots need to be able to function with no human intervention?

Yes, autonomy is the main focus of our project. In planetary exploration, specifically, when exploring underground caves, there is no, or very minimal, prior information about the environment. Further, when robots enter the cave, they typically lose communication with the surface and are on their own to accomplish the mission objectives.

As a result, autonomy is a crucial capability to enable such missions to accomplish mission goals with no human intervention when the robot is out of communication exploring previously unseen terrains and environments. To this end, JPL has been developing autonomy and AI solutions (called NeBula) acting as the robot brain, which is now being paired with Boston Dynamics Spot robots as the robot body.

Image Credits: Bryce Durbin/TechCrunch

A bit closer to Earth (as in, roughly 100 to 150 feet above our heads), Alphabet’s Wing announced this week that it’s approaching 100,000 drone deliveries two years after launching in the Brisbane-adjacent city of Logan, Australia. The announcement follows recent insight into Amazon’s struggles in the drone delivery space.

The company told TechCrunch,” I think we’ll launch new services in Australia, Finland and the United States in the next six months. The capabilities of the technology are probably ahead of the regulatory permissions right now.”

Image Credits: Wing

A closer look at some of those 100,000 deliveries:

  • 10,000 cups of coffee
  • 1,700 children’s snack packs
  • 1,200 hot chooks (roasted chicken, in Australian)
  • 2,700 sushi rolls
  • 1,000 loaves of bread

Image Credits: Coco

Speaking of food, LA-based Coco just raised $36 million for its delivery robots. The round brings its total funding up to $43 million. The UCLA spinout is currently piloting its 50-pound, remote-piloted robots in a variety of Los Angeles neighborhoods. The company tells TechCrunch:

We are currently operating in Santa Monica and in five different L.A neighborhoods. Later this year we are expanding into a number of other major U.S. cities. We have partnered with national restaurant brands like SBE (Umami Burger) and are actively scaling across many locations, and we are serving a wide range of family operated restaurants like Bangkok West Thai in Santa Monica and San Pedro Brewing Company in Los Angeles. We are out of the pilot phase and are launching with dozens of new merchants every day.

Image Credits: Spyce

Meanwhile, California-based fast casual salad chain Sweetgreens just acquired Spyce. Built by MIT alums, the company develops kitchen robotics, which it has rolled out in a pair of Boston-based restaurants. Sweetgreens eventually plans to implement these in some of its 120+ locations, though no timeline has been given as of yet.

In this week’s small-ray-of-sunshine-in-an-otherwise-horrifying-situation news, a team of young women roboticists managed to evacuate Kabul amid the Taliban takeover. The team has found refuge in Mexico on a 180-day humanitarian visa, with an option to extend their stay.

“From now on forward we will have opportunities for many more achievements in our lives, and thus be part of the fight for a better life,” team member Fatemah Qaderyan said at a press conference on their arrival in the country. “Although we are far from our homes, we will always be united and thanks to your help we will achieve it, thank you very much, we really appreciate having all our things here in Mexico with us.”

The team also made international headlines in 2017, as they entered the U.S. on a 10-day “parole,” in spite of the Trump administration’s executive order banning entry from predominately Muslim countries.

Before we go, a thought on the Tesla robot. Or, rather, a story. A few years ago, I was asked to be on a panel discussing robots for a group of people who weren’t really familiar with the field. That’s fine. There’s a lot to be said for getting outside your comfort zone. At the end, we opened things up to Q&A.

As is nearly always the case with these things, the first question — well, it wasn’t a question really. It was more of a laundry list of things the asker would like to see a robot do. She went on to describe a small drone that flies from surface to surface, cleaning different parts of the house. I told her it sounded great, and I’d love to see her invent it.

Point is, I think the vast majority of people outside of robotics have an entirely unrealistic idea of what’s possible with technology today. There’s a reason iRobot spent the better part of a decade banging its head against the wall, working out a robot that can vacuum floors. There’s also a reason that the Roomba is really the only semi-ubiquitous home robot. Always be wary of robots announced onstage as a press event.

I’m not saying a Tesla robot is impossible. I’m just saying we have to temper our expectations of what is. Sometimes you go in expecting a robot and get someone in a spandex onesie doing the Dougie:

Image Credits: Tesla

Salad chain Sweetgreen buys kitchen robotics startup Spyce

By Brian Heater

Like so many other aspects of the robotics world, the pandemic has dramatically accelerated interest in the automated kitchen. After all, the food and restaurant industry was deemed essential amid global shutdowns, but finding kitchen staff proved a problem for many, especially early on when questions remained around COVID’s transmission.

This week, California-based fast casual salad chain Sweetgreen announced plans to go all in on automation with the acquisition of Spyce. Founded in 2015, the Boston-based startup started making waves a few years back as a spinout of MIT mechanical engineering students. First serving up food at the school’s dining hall, the team ultimately opened a pair of automated restaurants in the Boston area. The startup notes, “our Spyce restaurants will stay open at this time.”

Sweetgreen plans to eventually incorporate Spyce’s technology into its restaurants. It will likely take some time to scale up to the needs of the chain, which currently operates more than 120 locations across the U.S.

Image Credits: Spyce

“We built Sweetgreen to connect more people to real food and create healthy fast food at scale for the next generation, and Spyce has built state-of-the-art technology that perfectly aligns with that vision,” Sweetgreen CEO and co-founder Jonathan Neman said in a statement. “By joining forces with their best-in-class team, we will be able to elevate our team member experience, provide a more consistent customer experience and bring real food to more communities.”

Like pizza, salads are a clear target for early food automation. They’re both popular and relatively straightforward to automate — essentially mixing a bunch of ingredients from different chutes into a bowl.

Sweetgreen is quick to note that the plan isn’t to replace employees outright, however.

“[T]eam members will be able to focus more on preparation and hospitality moments, while having the opportunity to work with state-of-the-art technology,” the company writes. “Invest more in training and development to support team members to become Head Coaches. Interested team members will be able to develop technology-facing skills to operate and maintain Spyce technology.”

The deal is expected to close in Q3. Terms were not disclosed.

Last-mile robotic delivery firm Coco raises $36M

By Brian Heater

Los Angeles delivery robot startup Coco this week has announced $36 million in funding. The Series A was led by Sam Altman, Silicon Valley Bank and Founders Fund, with participation from Sam Nazarian, Ellen Chen and Mario Del Pero. It brings the company’s total funding up to around $43 million.

“I strongly believe the delivery service industry in its current state is massively under-serving merchants. We have an enormous opportunity to create a better experience for hundreds of thousands of merchants and their customers, today,” co0founder and CEO Zach Rash said in a release. “This is not a research program experimenting with technology to be productized at some unknown point in the future.”

Image Credits: Coco

The UCLA spinout, formerly known as Cyan Robotics, is operating in a crowded field that includes names like Starship, Nuro and UC Berkeley alum Kiwibot. Rather than pushing for full autonomy, Coco’s solution utilizes remote drivers (which are a more popular solution than many companies care to admit).

Coco is still young, having launched in February 2020. The company currently has a headcount of 120, with plans to “grow to over 1,000” by end of year, as the pandemic continues to fuel additional interest in robotic deliveries. The new funding will also go toward hardware and additional city launches.

Coco says it has been able to operate with a 97% on-time rate, while reducing delivery times for its clients by around 30%. The company lacks a massive partner like Nuro’s work with Domino’s, though California-based Umami Burger is probably the largest on a list of 18 restaurant partners currently listed on Coco’s site.

“We are currently operating in Santa Monica and in five different L.A neighborhoods,” the company tells TechCrunch.  “Later this year we are expanding into a number of other major U.S. cities. We have partnered with national restaurant brands like SBE (Umami Burger) and are actively scaling across many locations, and we are serving a wide range of family operated restaurants like Bangkok West Thai in Santa Monica and San Pedro Brewing Company in Los Angeles. We are out of the pilot phase and are launching with dozens of new merchants every day.”

Actuator: Stop making sense

By Brian Heater

First of all, we’ve got a fancy new name. While “Robotics Roundup” was nothing if not very technically accurate, it lacked the kind of panache one ought to strive for when rounding up robotics. Actuator, on the other hand — that’s a mover and shaker.

It’s a name you can take to the bank (or at least run by the legal department for clearance). To mark this momentous occasion, we employed our resident graphic design genius Bryce to sketch up something befitting our rebrand.

We’re also using the opportunity to announce that Actuator will be coming soon to an inbox near you as a free TechCrunch newsletter. All of this fun change seems extra fitting, given that this happens to be the 25th edition of the roundup. You can find all of the older updates under our Actuator tag if you want to catch up.

If you’ve been following for a while, you’ve got the gist of what the newsletter is about: a digestible look into the week’s robotics news. We cover all of the startups making waves and the big companies impacting the industry, along with the most fascinating updates in the world of robotic research, as well as dives into labor concerns and various ethical issues stemming from automation and AI.

If all of that sounds good, you can sign up here to get Actuator in your inbox as soon as the first issue hits. I’m told you may have to prove you’re not a robot, so apologies in advance to all of the robots reading this. But hey, if you’ve gotten this far, you’ll figure it out.

Image Credits: Intel

Following an earlier report from CRN, Intel has since confirmed with TechCrunch that it will be winding down its 3D imaging platform, RealSense. It’s always a shame to see these sorts of forward-looking initiatives go away. And certainly Intel has been leaning pretty heavily on the division as a leading indicator of its efforts to remain relevant as the industry evolves.

Over the years, we’ve covered RealSense’s involvement in drones, robotics and AR/VR. In June of last year, we covered the platform’s embrace of 5G connectivity.

Image Credits: Intel

“We are winding down our RealSense business and transitioning our computer vision talent, technology and products to focus on advancing innovative technologies that better support our core businesses and IDM 2.0 strategy,” the company said in a statement offered to TechCrunch. “We will continue to meet our commitments to our current customers and are working with our employees and customers to ensure a smooth transition.

Translation: The company is choosing to focus its core competency. IDM 2.0 refers specifically to the new chipmaking strategy into which the company is pumping $20 billion. Understandable, but it’s always hopeful to see big companies like Intel, Nvidia and Qualcomm really go all in on such forward-facing technologies.

Boston Dynamics, meanwhile, made news this week, ostensibly for another slick viral video, this one featuring the Hyundai-owned company’s humanoid Atlas robot. By now we’re all well aware of the fact that the company makes impressive robots and highly effective YouTube videos that launch a million Black Mirror and Terminator jokes on Twitter.

I’ve seen Atlas do some really impressive stuff in person at BD’s headquarters, and I’ve got a pretty good idea of what it’s currently capable of. So, while Atlas is extremely cool, I didn’t find the recent parkour video especially shocking. What did catch me off guard, however, was the fact that the company also used the opportunity to essentially publish some outtakes from the film.

Image Credits: Boston Dynamics

A six-minute, behind-the-scenes video featured a montage of Atlas falling on its face. Like any great skateboarding video, there are a few gratuitous shots included that demonstrate that, regardless of how advanced the system is, there are still going to be some face-planting, gasket-blowing falls that leave its chest scuffed in a pool of its own fluid. The company notes:

During filming, Atlas gets the vault right about half of the time. On the other runs, Atlas makes it over the barrier, but loses its balance and falls backward, and the engineers look to the logs to see if they can find opportunities for on-the-fly adjustments.

"We'll be singing / when we're winning." 🔊pic.twitter.com/51DYD1Avvg

— Brian Heater (@bheater) August 17, 2021

That’s probably enough news of shuttered divisions and bodily robot harm for this week. A couple of fundraising rounds are worth noting.

First is Rapid Robotics, which has been on a fundraising tear of late. The new $36.7 million Series B values the manufacturing robotics company at $192.5 million and marks its third(!) fundraising round in a year that started with a seed raise.

Image Credits: Rapid Robotics

CEO Jordan Kretchmer cites pandemic-fueled manufacturing bottlenecks as a big source of interest in the company:

We hear a lot about the semiconductor shortage, but that’s just the tip of the iceberg. Contract manufacturers can’t produce gaskets, vials, labels — you name it. I’ve seen cases where the inability to produce a single piece of U-shaped black plastic brought an entire auto line to a halt

Image Credits: Diamond Age

Rapid will be making its robotic systems available through the increasingly popular RaaS (robotics as a service) model also being employed by Diamond Age. The fellow Bay Area-based firm announced its own $8 million seed round this morning for an intriguing mix of robotics and 3D printing designed at speeding up house construction. The company is still in its early stages, but it claims its technology can dramatically reduce the need for manual labor and shrink house construction time from nine months to 30 days.

Image Credits: Picnic

Following its own recent funding back in May, Picnic this week announced that it’s finally selling its modular robotic pizza maker. Pizza is, of course, a popular target for food robotics companies, because Americans eat a ton of it — reportedly 100 acres a day, as of 2015. It’s also relatively uniformly constructed as far as self-contained meals go, and is therefore easier to automate.

Nuro-validation test

Nuro team on test track during early validation in Arizona, before first-ever public road deployment in Arizona. Image Credits: Nuro

And speaking of pizza robots, before we leave you this week, a note to check out the EC-1 on Nuro. Here’s a fun anecdote from Domino’s chief innovation officer that seems to ring true across the robotic spectrum:

One of the things we laugh about is how customers constantly talk to the bot. It’s almost like they think it’s ‘Knight Rider.’ It’s very common for customers to thank it or say goodbye, which is great because that indicates we’re creating an engaging experience that they’re not frustrated by.

Diamond Age raises $8M to speed up home construction with 3D printing and robot arms

By Brian Heater

Bay Area-based Diamond Age this week announced that it has raised $8 million. The seed round is led by Prime Movers Lab and Alpaca VC and features a slew of additional firms, including Dolby Family Ventures, Calm Ventures, Gaingels, Towerview Ventures, GFA Venture Partners and Suffolk Construction.

The startup looks to put a slew of key emerging technologies to work in service of building houses with fewer workers in a significantly truncated time frame. Diamond Age claims that when, fully realized, its tech will be able to reduce manual human labor by 55% and shrink the construction time on a single family home from nine months to 30 days. Part of this funding will go toward putting the processes in place to construct a 1,100-square-foot “demonstration house” as proof of concept.

“We need to build high-quality affordable single-family homes for the next generation striving for the American dream,” co-founder and CEO Jack Oslan said in a release, “and the only way we can solve this problem is with automation.”

Specifically, the company relies on robotic and 3D printing solutions. The former involves a set of 26 different robotic arm attachments to assist with the construction. That tech is coupled with a gantry-based 3D printing technology designed to construct interior and exterior walls for the structure.

Specifically, the company is looking to target the housing crunch in its Bay Area backyard. The systems will be available to construction companies through a RaaS (robotics as a service) rental model. Pricing specifics for the system have not been revealed.

Bedrock modernizes seafloor mapping with autonomous sub and cloud-based data

By Devin Coldewey

The push for renewable energy has brought offshore wind power to the forefront of many an energy company’s agenda, and that means taking a very close look at the ocean floor where the installations are to go. Fortunately Bedrock is here to drag that mapping process into the 21st century with its autonomous underwater vehicle and modern cloud-based data service.

The company aims to replace the standard “big ship with a big sonar” approach with a faster, smarter, more modern service, letting companies spin up regular super-accurate seafloor imagery as easily as they might spin up a few servers to host their website.

“We believe we’re the first cloud-native platform for seafloor data,” said Anthony DiMare, CEO and cofounder (with CTO Charlie Chiau) of Bedrock. “This is a big data problem — how would you design the systems to support that solution? We make it a modern data service, instead of like a huge marine operation — you’re not tied to this massive piece of infrastructure floating in the water. Everything from the way we move sonars around the ocean to the way we deliver the data to engineers has been rethought.”

The product Bedrock provides customers is high-resolution maps of the seafloor, made available via Mosaic, a familiar web service that does all the analysis and hosting for you — a big step forward for an industry where “data migration” still means “shipping a box of hard drives.”

Normally, DiMare explained, this data was collected, processed, and stored on the ships themselves. Since they were designed to do everything from harbor inspections to deep sea surveys, they couldn’t count on having a decent internet connection, and the data is useless in its raw form. Like any other bulky data, it needs to be visualized and put in context.

Example of data in the Mosaic system from Bedrock, showing a map and trails of data points.

Image Credits: Bedrock

“These datasets are extremely large, tens of terabytes in size,” said DiMare. “Typical cloud systems aren’t the best way to manage 20,000 sonar files.”

The current market is more focused on detailed, near-shore data than the deep sea, since there’s a crush to take part in the growing wind energy market. This means that data is collected much closer to ordinary internet infrastructure and can be handed off for cloud-based processing and storage more easily than before. That in turn means the data can be processed and provided faster, just in time for demand to take off.

As DiMare explained, while there may have been a seafloor survey done in the last couple decades of a potential installation site, that’s only the first step. An initial mapping pass might have be made to confirm the years-old maps and add detail, then another for permitting, for environmental assessments, engineering, construction, and regular inspections. If this could be done with a turnkey automated process that produced even better results than crewed ships for less money, it’s a huge win for customers relying on old methods. And if the industry grows as expected to require more active monitoring of the seafloor along every U.S. coast, it’s a win for Bedrock as well, naturally.

CG render of the AUV.

Image Credits: Bedrock

To make this all happen, of course, you need a craft that can collect the data in the first place. “The AUV is a piece of technology we built solely to enable a data product,” said DiMare, but noted that, originally, “we didn’t want to do this.”

“We started to spec out what it looked like to use an off the shelf system,” he explained. “But if you want to build a hyper-scalable, very efficient system to get the best cost per square meter, you need a very specific set of features, certain sonars, the compute stack… by the time we listed all those we basically had a self-designed system. It’s faster, it’s more operationally flexible, you get better data quality, and you can do it more reliably.”

And amazingly, it doesn’t even need a boat — you can grab it from the back of a van and launch it from a pier or beach.

“From the very beginning one of the restrictions we put on ourselves was ‘no boats.’ And we need to be able to fly with this thing. That totally changed our approach,” said DiMare.

View of the AUV on a beach

Image Credits: Bedrock

The AUV packs a lot into a small package, and while the sensor loadout is variable depending on the job, one aspect that defines the craft is its high-frequency sonar.

Sonars operate in a wide range of frequencies, from the hundreds to the hundreds of thousands of hertz. Unfortunately that means that ocean-dwelling creatures, many of which can hear in that range, are inundated with background noise, sometimes to the point where it’s harmful or deters them from entering an area. Sonar operating about 200 kHz is safe for animals, but the high frequency means the signal attenuates more quickly, reducing the range to 50-75 meters.

That’s obviously worthless for a ship floating on the surface — much of what it needs to map is more than 75 meters deep. But if you could make a craft that always stayed within 50 meters of the seabed, it’s full of benefits. And that’s exactly what Bedrock’s AUV is designed to do.

The increased frequency of the sonar also means increased detail, so the picture its instruments paint is better than what you’d get with a larger wave. And because it’s safe to use around animals, you can skip the (very necessary but time-consuming) red tape at wildlife authorities. Better, faster, cheaper, and safer is a hell of a pitch.

Today marks the official launch of Mosaic, and to promote adoption Bedrock is offering 50 gigs of free storage — of any kind of compatible map data, since the platform is format-agnostic.

There’s a ton of data out there that’s technically “public” but is nevertheless very difficult to find and use. It may be a low-detail survey from two decades ago, or a hyper-specific scan of an area investigated by a research group, but if it were all in one place it would probably be a lot more useful, DiMare said.

“Ultimately we want to get where we can do the whole ocean on a yearly basis,” he concluded. “So we’ve got a lot of work to do.”

Is the Robot-Filled Future of Farming a Nightmare or Utopia?

By Katrina Miller
A new paper argues that the rise of artificial intelligence in agriculture could be the best—or worst—innovation for our environment.

Peer into the eyes of Cyberdog

By Brian Heater

When someone mentioned to me that Xiaomi was launching its own “robot dog,” my mind immediately went to Sony’s Aibo. And honestly, it would have been difficult to be more wrong. Now that the news has been out for a few days, the company’s heard all of your bad Black Mirror jokes, don’t worry.

And, honestly, the Chinese hardware maker didn’t do itself any favors with the design here. Boston Dynamics has done a lot to imbue its quadrupedal robots with personality, through design language and viral videos of Spot and company busting a move to the Dirty Dancing soundtrack.

With Cyberdog, however, Xiaomi’s design team clearly just leaned in and went full-on Robocop (and the Bladerunner pastiche doesn’t help) . I receive a deluge of Metalhead gifs every time I post something about Boston Dynamics — seriously, I’m using Cyberdog as the lead image on this post, just so you can see what I mean. Go check the replies on Twitter. I’ll wait.

Image Credits: Xiaomi

Xiaomi is, of course, far from the first company to release a Spot-like quadrupedal robot. There are a number of companies competing in that space, including ANYmal and Ghost Robotics. For its part, Xiaomi is looking to put a developer spin on the category. Per the Mi blog:

CyberDog is Xiaomi’s first foray into quadruped robotics for the open source community and developers worldwide. Robotics enthusiasts interested in CyberDog can compete or co-create with other like-minded Xiaomi Fans, together propelling the development and potential of quadruped robots.

Image Credits: Xiaomi

The robot is powered by Nvidia’s Jetson Xavier NX platform, coupled with 11-built in sensors, including cameras, touch, GPS and more. The company will be release 1,000 of the robots, price at roughly $1,540 — a fraction of the cost of the advanced Spot system. The robot is also a fraction of the size of Boston Dynamics’ quadruped. And while there are superficial similarities the project really couldn’t be more different.

Xiaomi’s entry into robotics is more about building hardware for Nvidia’s platform. It’s a (relatively) inexpensive way for people to get a hang of programming and, perhaps, protoyping robotics. The likely limited functionality — and availability — are pretty clear indications that that the company’s not trying to put a Cyberdog in every home just yet.

Bear Flag Robotics

A sizable acquisition this week, John Deere announced plans to buy Bear Flag Robotics for $250 million. We’ve been following Bear Flag since it was a member of the YC cohort. The deal seems like a good outcome for both parties. Bear Flag gets a lot of resources from an agricultural giant like John Deere and Deere gets to step another foot into the world of cutting-edge tech with an autonomous tractor startup.

Says co-founder and CEO Igino Cafiero:

One of the biggest challenges farmers face today is the availability of skilled labor to execute time-sensitive operations that impact farming outcomes. Autonomy offers a safe and productive alternative to address that challenge head on. Bear Flag’s mission to increase global food production and reduce the cost of growing food through machine automation is aligned with Deere’s and we’re excited to join the Deere team to bring autonomy to more farms.

Image Credits: Kiwibot

Another startup we’ve been following since its early days, Kiwibot is seeing expansion to a significant number of campuses. In spite of campus shutdowns last year, the Berkeley-based company is actually seeing something of a boom due to the pandemic. COO Diego Varela Prada tells TechCrunch:

We have a procedure to disinfect the bots between orders. If you’re a student and you don’t want to mix into large crowds, I think it’s much safer to order food through Kiwibot and have it delivered to the library or your dorm.

We’ve written about Lidar company Aeva a few times over the years, including last November, when it announced plans to go public via SPAC. This week, the company announced a deal with Nikon that takes it beyond its existing automotive applications. The company says there are a slew of potential applications, though the chip is still about four years away from production. Fields include, “consumer electronics, consumer health, industrial robotics, and security.”

A whole bunch of robots are making their way to Florida late next year, courtesy of Amazon. The company announced this week that it has chosen Tallahassee (birthplace of T-Pain and objectively the best Mountain Goats album) as the home of its next fulfillment center. The company plans to add to its massive arm of warehouse robots for the 630,000-square-foot space, along with 1,000 human jobs.

Image Credits: Berkshire Grey

FedEx, meanwhile, has implemented Berkshire Grey robotics at a shipping facility in Queens (the best borough). The systems will identity, pick, sort, collected and containerize primarily small packages like polybags, tubes and padded mailers. The systems are set to roll out to additional locations, including Las Vegas and Columbus, Ohio. Says B.G.,

This technology has been developed and installed as a direct response to the exponential growth of e-commerce, which has accelerated the demand for reliable automated solutions throughout all stages of the supply chain. FedEx Ground believes that continued innovation and automation will improve safety, efficiency, and productivity for its team members as they continue to keep the e-commerce supply chain moving.

Image Credits: Hyphen

Here’s a new company in the food space worth keeping an eye on. Formerly known as Ono Food Co. (then a food truck company), SF-based Hyphen has come out of stealth with the announcement of its Makeline automated meal platform. The company says the system is able to create up to 350 meals an hour, with the aid of a single staff member.

“[W]e really see ourselves like Shopify,” CEO Stephen Klein said in a release, “but instead of enabling merchants to compete with the likes of Amazon, we’re enabling restaurants to compete with the likes of DoorDash as well as other services and ghost kitchens that have decided to compete with their own customers by offering their own food brands.”

The platform is set to start rolling out this winter with plans for 300 locations in New York City, San Francisco, Los Angeles, Seattle and Phoenix.

Amazon’s $1.5 billion US air cargo hub is open for business

By Kirsten Korosec

Amazon’s $1.5 billion air cargo hub in Northern Kentucky opened Wednesday, the latest effort by the e-commerce giant to connect a network of 40 sites and control all aspects of delivery as demand for speed and convenience accelerates.

The Amazon Air Hub operations, located at the Cincinnati/Northern Kentucky International Airport, will be the center of its U.S. cargo network. The hub opened after more than four years of planning and construction. Amazon said the U.S. hub will eventually operate a dozen flights per day and process millions of packages every week.

The hub is comprised of an 800,000-square-foot sortation building located on a 600-acre campus that includes seven buildings, a new ramp for aircraft parking and a multistory vehicle parking structure.

Amazon said that eventually more than 2,000 people will be employed there. The air hub will also rely on robotics technology, specifically robotic arms to move and sort packages and mobile drive units to transport packages across the building.

Amazon Air launched in 2016 and has grown into a network of more than 40 locations. Last year, Amazon Air launched its European air hub at Germany’s Leipzig/Halle Airport, a 215,000-square-foot facility that hosts two Amazon-branded Boeing 737-800 aircraft.

Amazon Air also has regional air hubs at airports in Texas, Puerto Rico and Florida in the U.S., and plans to expand to San Bernardino International Airport in California and Cincinnati/Northern Kentucky International Airport in 2021.

Starship Technologies is bringing food delivery robots to four more US college campuses this year

By Rebecca Bellan

Starship Technologies, an autonomous delivery services company, announced that it will begin delivery service on four additional college campuses this fall, adding to the 20 campuses on which it already operates.

The University of Illinois Chicago (UIC), the University of Kentucky (UK), the University of Nevada, Reno (UNR) and Embry-Riddle Aeronautical University’s Daytona Beach, Florida campus will all be graced with the Estonian-born company’s little six-wheeled, zero-emissions delivery robots.

This announcement comes the same day as Kiwibot, another autonomous sidewalk delivery robot company, has partnered with hospitality giant Sodexo to bring food delivery to college campuses. Whereas Kiwibot will focus more on delivering food from dining halls and university stores, it looks like Starship will work with on-campus merchants like Starbucks, Panda Express and Panera Bread, among others. Despite the different approaches, the outcome is the same: Delivery companies are preparing for a more “normal” school year, even though the Delta variant continues to ramp up case loads. That could either be a blessing or a problem for the Starships and Kiwibots of the world. On the one hand, more ‘Rona means more students staying inside and avoiding other humans. On the other, it could also mean school shutdowns and a bunch of useless bots.

“We see the Starship robots as an important part of safely bringing students back to campus,” said Dean Kennedy, executive director of residential life, housing and food services at UNR, in a statement. “Everyone wants to resume in-person classes and be back on campus so we’re doing everything we can to make sure it’s done responsibly. The robots offer several advantages – they make social distancing easier, they are convenient, the students we have spoken with love this idea and they continue our heritage of being an innovative campus.”

UIC will have 25 Starship robots and UNR and Embry-Riddle will get 20 robots each, all of which add to Starship’s fleet of over 1,000 delivery robots. The company says it has completed more than 1.5 million rides since its founding in 2014. It has raised a total of $102 million, including its recent $17 million funding round.

“We’ve worked hard to become a trusted and integrated partner on our campus communities and that hard work has paid off,” said Alastair Westgarth, CEO of Starship, in a statement. “We are continuing to add new schools every semester, with more to be announced this fall. The students love the robots and the schools appreciate the ability to offer this service. We can’t wait to meet the students at each of these schools and look forward to hiring students on all of the campuses to give them real world experience working with robots and AI.”

Students and faculty will be able to download the Starship Food Delivery app to choose meals and then drop a pin where they want their delivery to be sent. They can track the robot or they can wait for an alert to go outside and meet the robot once it has arrived, where they can then unlock it through the app. Starship says it aims to train and hire students at local campuses who are interested in joining the team and learning more about autonomous technology.

Kiwibot partners with hospitality giant Sodexo to bring food delivery robots to more college campuses

By Rebecca Bellan

Kiwibot, the robotic sidewalk delivery startup, has announced a partnership with food services and facilities management giant Sodexo to bring its robots to U.S. college campuses. As of this month, students and faculty at New Mexico State University, Loyola Marymount University and Gonzaga University should have the option to order fresh meals via cute little robots from Sodexo-serviced locations on campuses.

This is not the first time Kiwibot is delivering food for over-caffeinated, hungover, exhausted college kids. Its robots, which are designed to look adorable and can move at around two miles per hour, were born at the University of Berkeley, California. There, the company was able to rack up 150,000 deliveries and prove its use case for expansion to other campuses, like the University of Denver and Stanford University. 

Sodexo provides the food for the cafeterias and dining halls of hundreds of colleges across the U.S., so this partnership could prove to be massively fruitful for Kiwibot. It’ll need the boost in order to keep up with its main competition, Starship Technologies, which completed one million delivery rides in January and operates in many college campuses, as well. 

Campuses are often a natural choice for startups in autonomous development. Not only are universities open to experimenting with new ideas, but given the unit economics on a campus, the revenue growth prospects are more favorable than working B2B in a city, says Diego Varela Prada, chief operating officer at Kiwibot. 

“Additionally, university campuses provide an advantage as they tend to be a more controlled environment than public streets in regards to things such as public infrastructure complexity and car traffic and congestion,” Prada told TechCrunch.

As part of the partnership, students will be able to use their meal plans for Kiwibot delivery through their Sodexo Bite+ app. Those without a meal plan can pay à la carte, including $2 flat fee plus 10% of the order amount. 

“We’re starting with 10 bots at Loyola, 10 bots at Gonzaga and 30 bots at New Mexico State, and that’s just the beginning,” said Prada. “We’re hoping to have many more. As a B2B business, we’re able to work very closely with our partner to increase the capacity of the bots as demand ramps up.”

Last year, many college campuses were forced to shut down amid the pandemic. Now, as school is about to be in session again, the U.S. is experiencing the highest caseload since February, averaging about 100,000 new cases every day. There’s no telling what kinds of shutdowns or lockdowns we’ll see on college campuses this year, but Prada sees Kiwibot providing an avenue for students to order food safely.

“We have a procedure to disinfect the bots between orders,” he said. “If you’re a student and you don’t want to mix into large crowds, I think it’s much safer to order food through Kiwibot and have it delivered to the library or your dorm.”

Prada says Kiwibot’s robots, which are in their fourth generation, are advancing to Level 4 autonomy, but are currently at Level 3. The Society of Automotive Engineers describes both Levels 3 and 4 as a full self-driving system, but with Level 3, a human operator may be required to take over if there’s an issue, whereas a Level 4 system is expected to handle all driving on its own. 

“We have a feature that’s called corner-to-corner, and so what that does is it captures data around an indoor space and feeds that into an algorithm that basically makes decisions for the bot,” said Prada. “It handles the navigation for the robot in between high complexity situations, like cars, people, pets, little kids, people that work on campus. So we are not in a position yet to let the bot go on the campus on its own. Our remote operators have a feature where they switch to corner-to-corner and if the bot senses, for example, a street pass, then the remote operator or supervisor will take over.”

The engineers at Kiwibot are working on getting a bot to navigate indoor-to-outdoor and outdoor-to-indoor, so that it can pick up orders in a kitchen, go outside to deliver it and then navigate inside a building so that it can deliver an order to someone’s room or desk.

John Deere Doubles Down on Silicon Valley and Robots

By Will Knight
The farm-equipment giant is buying Bear Flag Robotics, which makes autonomous tractors, marking its second big tech buy in four years.

The Station: The script for Elon Musk’s Loop drivers, Redwood snags $700M and a chat with Kodiak Robotics’ co-founder

By Kirsten Korosec

The Station is a weekly newsletter dedicated to all things transportation. Sign up here — just click The Station — to receive it every weekend in your inbox.

Hello readers: Welcome to The Station, your central hub for all past, present and future means of moving people and packages from Point A to Point B.

In case you missed it, our scoop machine Mark Harris was at it again. This time, he found some interesting and entertaining documents related to Elon Musk’s underground Loop system in Las Vegas received via a Freedom of Information Act. Among the treasure is a “ride script” that instructs drivers for the Loop system to bypass passengers’ questions about how long they have been driving for the company, declare ignorance about crashes, and shut down conversations about Musk himself.

The takeaway: the script shows just how serious The Boring Company, which built and operates the system, is about controlling the public image of the new system, its technology and especially Musk.

Importantly, the documents confirm that Autopilot, the advanced driver assistance system in the Tesla vehicles used in the Loop system, must be disabled.

As always, you can email me at kirsten.korosec@techcrunch.com to share thoughts, criticisms, offer up opinions or tips. You can also send a direct message to me at Twitter — @kirstenkorosec.

Micromobbin’

This is a step outside the norm of what I usually think of when I think of micromobility (you’ll see what I did there in a second), but this week I wrote about a new in-shoe navigation system that helps the visually impaired walk around town.

Ashirase, as both the system and the name of the company is called, involves attaching a three-dimensional vibration device, including a motion sensor, inside a pair of shoes. This bit of hardware is connected to a smartphone app that someone with low vision can use to enter their destination. Vibrations in the front part of the shoe give the cue to walk straight, and vibrations on the left and right cue the user to make a left or right turn. The aim is to free up the hands while walking to use a cane and allow the walker to put more of their full attention on audio signals in the environment, thus making their commutes a bit more intuitive and their lives more independent.

It’s a really interesting bit of tech because it uses a similar stack to what we’re seeing in autonomous driving and advanced driver assistance systems. Which makes sense because that’s the founder’s background. Wataru Chino worked in Honda’s EV motor control and automated driving systems departments since 2009. His startup is a product of Honda’s incubator, Ignition, that features original technology, ideas, and designs of Honda associates with the goal of solving social issues and going beyond the existing Honda business.

Accessibility: We love to see it

Cabify recently announced a new feature that makes its rideshare service more accessible to the elderly, people with partial visual impairment and people with cognitive disabilities. The feature provides voice notifications to alert the user when a driver is on their way or has just arrived, when the ride starts, when a stop has been reached, when a message has come into the app’s chat, etc.

The notification makes use of a text-to-speech functionality that Android and iOS phones have.

“Apple and Google operating systems allow us to pronounce sentences with the system’s voice but we have developed the text and established the situations where we inform and draw the user’s attention,” a Cabify spokesperson told me.

Lime’s push for world domination

And we’re back with the latest on Lime’s plans to take over the world, one electric scooter at a time. The micromobility goliath has announced an integration with the Moovit transit planning app. From Monday onwards, Moovit users in 117 cities across 20 countries will see Lime’s electric scooters, bikes and mopeds show up as an option for travel, either as the whole journey or as part of a multi-modal journey. This news follows a trend we’re seeing as cities start to see micromobility companies as less of a public nuisance and more of a public solution, particularly for first- and last-mile travel. Integrating with Moovit, an app that’s solely focused on public transportation, is a move that helps in the long run creating a broader transportation ecosystem.

New whips

Espin released its limited edition fixie style e-bike called the Aero. It’s just the thing for Seattle hipsters, particularly ones with a stick-and-poke bike tattoo. The bike frame is just as sleek as you’d expect from a single gear bike, all clean lines and comes in either a forest green or a smoke gray. The Aero can reach top speeds of 20 mph and can hit 30 miles on a single charge. Best of all, it doesn’t break the bank at $1,399.

Splach, which normally makes e-scooters and e-bikes, has come out with something it’s calling the Transformer. I truly don’t know how to categorize it but it looks like a lot of fun to ride. The company is calling the light-duty e-vehicle a “mini-moto Robust scooter specialized for rugged terrains.” It looks like a dirt bike has been sized way down and given a long neck so you can stand on it and still steer it. It also looks like it would indeed do well on rugged terrains, based on videos of people shredding down dirt paths. Splach used Indiegogo to fund the thing, and said it reached its goal within an hour.

Deal of the week

money the station

Get ready to hear a lot more about supply chain constraints around batteries with virtually every automaker shouting out pledges to shift their entire portfolio away from internal combustion engines and towards electric powertrains.

Cell producers need access to the raw materials like nickel that are needed to make batteries. Mining those materials is the most common means, but that isn’t sustainable (and I’m not just talking about the environmental toll). JB Straubel, who is best known as the former Tesla co-founder and longtime CTO, is tackling the supply chain issue through his startup Redwood Materials. The battery recycling company is aiming to create a circular supply chain. This closed-loop system, Straubel says, will be essential if the world’s battery cell producers hope to have the supply needed for consumer electronics and the coming wave of electric vehicles.

High-profile investors like Amazon, funds managed by T. Rowe and Bill Gates’ Breakthrough Ventures fund recognize the opportunity and have injected $700M in fresh capital into Redwood Materials. This is comically large compared to the startup’s last raise of $40 million. And sources tell me that this pushes Redwood Material’s valuation to $3.7 billion.

I interviewed Straubel about the raise and what struck me was how aggressively he wants to scale; he is treating this issue as if there is no time to lose — and he’s not wrong.

Other deals that got my attention this week …

Clarios, the maker of low-voltage vehicle batteries, postponed its IPO, citing market volatility, Bloomberg reported. the Milwaukee area-based company backed by Brookfield Asset Management had filed to raise $1.7 billion by offering 88.1 million shares at a price range of $17 to $21.

Fisker, the electric vehicle startup turned publicly traded company via a SPAC, has turned investor to support EV charging company Allego. Fisker said it is investing $10 million in private-investment-in-public equity (PIPE) funding for the merger of Allego and special purpose acquisition company Spartan Acquisition Corp III. The merger puts Allego at a pro forma equity value of $3.14 billion.

Flock, which went from providing drone insurance to commercial vehicle insurance, raised $17 million in a Series A funding led by Social Capital, the investment vehicle run by Chamath Palihapitiya, best known as a SPAC investor and chairman of Virgin Galactic. Flock’s existing investors Anthemis and Dig Ventures also participated. This round brings Flock’s total funding to $22 million. Justin Saslaw (Social Capital’s fintech partner) joins Flock’s board of directors, as does Ross Mason (founder of Dig Ventures and MuleSoft).

HappyFresh, the on-demand grocery app based in Indonesia, raised $65 million in a Series D round led by Naver Financial Corporation and Gafina B.V., with participation from STIC, LB and Mirae Asset Indonesia and Singapore. It also included returning investors Mirae-Asset Naver Asia Growth Fund and Z Venture Capital. The company’s previous round of funding was a $20 million Series C announced in April 2019.

Lordstown Motors got a lifeline from a hedge fund managed by investment firm Yorkville Advisors about five weeks after the automaker issued a warning that it might not have enough funds to bring its electric pickup truck to market. The hedge fund agreed to buy $400 million worth of shares over a three-year period, according to a regulatory filing.

Merqueo, the on-demand delivery service that operates in Latin America, raised $50 million in a Series C round of funding co-led by IDC Ventures, Digital Bridge and IDB Invest. MGM Innova Group, Celtic House Venture Partners, Palm Drive Capital and previous shareholders also participated. The financing brings the Bogota, Colombia-based startup’s total raised to $85 million since its 2017 inception.

Niron Magnetics, a company developing permanent magnets free of rare earths, raised $21.3 million in new financing from the Volvo Cars Tech Fund and Volta Energy Technologies, which joined existing investors Anzu Partners and the University of Minnesota. Niron will use the funding to build its pilot production facility in Minnesota.

Onto, the EV car subscription company raised $175 million in a combined equity and debt Series B round. The equity piece was led by Swedish VC Alfvén & Didrikson. British investment company Pollen Street Capital provided the senior-secured asset-backed debt facility. The company, which has raised a total of $245 million, says it plans to double its fleet size every three to six months and that any new vehicles will be used as collateral. Onto did not disclose how much of the round came from equity versus debt.

Zūm, a student transportation startup, was awarded a five-year $150 million contract to modernize San Francisco Unified School District transport service throughout the district. Zūm, which already operates its rideshare-meets-bus service in Oakland, much of Southern California, Seattle, Chicago and Dallas, will be responsible for handling day-to-day operations, transporting 3,500 students across 150 school campuses starting this fall semester.

A little bird

blinky cat bird green

I hear things. But I’m not selfish. Let me share!

You might have missed my article late Friday about Argo AI landing a permit in California that will allow the company to give people free rides in its self-driving vehicles on the state’s public roads.

Tl;dr: The California Public Utilities Commission issued Argo the so-called Drivered AV pilot permit, which is part of the state’s Autonomous Vehicle Passenger Service pilot. This puts Argo in a small and growing group of companies seeking to expand beyond traditional AV testing — a signal that the industry, or at least some companies, are preparing for commercial operations.

Regulatory hurdles remain and don’t expect Argo to be offering and charging for “driverless” rides anytime soon. But progress is being made and I would expect the company to secure the next permit — in a long line of them — later this year.

Argo has never officially indicated what city it is targeting for a robotaxi service in California. The company has been testing its autonomous vehicle technology in Ford vehicles around Palo Alto since 2019. Today, the company’s test fleet in California is about one dozen self-driving test vehicles. It also has autonomous test vehicles in Miami, Austin, Washington D.C., Pittsburgh and Detroit. (In July, Argo and Ford announced plans to launch at least 1,000 self-driving vehicles on Lyft’s ride-hailing network in a number of cities over the next five years, starting with Miami and Austin.)

I’m hearing from some sources familiar with Argo’s strategy for California that we should look south of the Bay Area. Way south.

The city that jumps to mind is San Diego. Some AV companies are already playing around the Irvine area and Los Angeles seems too unwieldy. Plus, Ford already has a footprint in San Diego. The automaker partnered way back in 2017 with AT&T, Nokia and Qualcomm Technologies to test Cellular vehicle-to-everything (CV2X) at the San Diego Regional Proving Ground with the support of the San Diego Association of Governments, Caltrans, the city of Chula Vista, and intelligent transportation solutions provider McCain. The upshot of these trials? To improve traffic efficiency, vehicle safety and “support a path towards autonomous vehicles.”

Policy corner

the-station-delivery

Hi everyone. Let’s dive into two key pieces of proposed legislation this week: the infrastructure bill and the tailpipe emissions standards.

After months of negotiations, U.S. senators have finally settled on a $550 billion infrastructure package that includes investments in roads, bridges, broadband and more. The bill would provide $7.5 billion to electrify buses and ferries, including school buses, and $7.5 billion to build out a national network of public EV charging stations. Subsequent statements on the bill from the White House say directly that the EV investments are intended to keep the U.S. competitive on the world stage: “U.S. market share of plug-in electric vehicle (EV) sales is only one-third the size of the Chinese EV market. The President believes that must change.”

The budget is just a fraction of the $2.25 trillion bill President Joe Biden originally introduced in March. That version of the bill earmarked billions more for transportation electrification, especially in rebates and incentives to get consumers buying more EVs. The bill is still with the Senate for final approval. Then it will head to the House before finally ending up on Biden’s desk.

The Environmental Protection Agency and the Department of Transportation have proposed rules that would beef up tailpipe emissions standards, which had been rolled back under President Donald Trump. The rules would be identical to the agreement the state of California reached with Ford, VW, Honda, BMW and Volvo in 2019, the AP reported. If approved, the rules would apply starting with model year 2023 vehicles.

The aim is to cut carbon emissions from transportation and encourage more people to buy hybrid and electric. But many environmental groups like the Sierra Club — plus some EV automakers — don’t think they go far enough.

“This draft proposal would drive us in the right direction after several years in reverse–but slowly getting back on track is not enough,” Chris Nevers, senior director of environmental policy at Rivian, told TechCrunch. EPA and NHTSA must maximize the stringency of the program beyond the voluntary deal and account for current and future developments in vehicle electrification.

One more thing that caught my eye this week…The Washington Post reported that Biden and a group of automakers are negotiating for the latter group to make a “formal pledge” to have at least 40% of all vehicles sold in 2030 to be electric. The article doesn’t specify which OEMs are part of the talks. However, it’s hard to imagine automakers signing onto anything — even a “voluntary pledge” — without some hefty federal spending to go along with it. We’ll have to see if the provisions in the infrastructure bill are enough.

— Aria Alamalhodaei

Notable reads and other tidbits

As per ushe, there was a ton of transportation news this week. Let’s dig in.

ADAS

Yep, ADAS gets its own section now in an effort to make it abundantly clear that advanced driver assistance systems are not self-driving cars. Never. Never ever.

New York Times’ Greg Bensinger weighs in on beta testing and Tesla in this opinion column.

Autonomous vehicles

Aurora co-founder and chief product officer Sterling Anderson put out a blog and a bunch of tweets to layout a blueprint for an autonomous ride-hailing business that will launch in late 2024 with partners Toyota and Uber. Aurora has spent the past year or so pushing its messaging on self-driving trucks, which the company says is its best and most viable first commercial product. Aurora never entirely ditched the robotaxi idea, but it was pretty quiet on the topic. Until now.

The blog comes about a week after competitor Argo AI and Ford announced a partnership with Lyft. While the timing might not be related, it does show that competition is heating up in both areas — robotaxis and self-driving trucks — with every AV company keen to show progress and deep partnerships.

TuSimple, the self-driving truck company that went public earlier this year, has partnered with Ryder as part of its plan to build out a freight network that will support its autonomous trucking operations. Ryder’s fleet maintenance facilities will act as terminals for TuSimple’s so-called AFN, or autonomous freight network.

Electric vehicles

Ford released Wednesday its second quarter earnings for 2021, which besides containing a surprise profit despite the ongoing chip shortage, revealed that its F-150 Lightning electric pickup has generated 120,000 preorders since its unveiling in May. Ford reported revenue of $26.8 billion, slightly below expectations, and net income of $561 million in the second quarter.

Lucid Group (formerly Lucid Motors) will be expanding its factory in Casa Grande, Arizona, by 2.7 million square feet, CEO Pete Rawlinson said just hours after the company officially went public with a $4.5 billion injection of capital. The company also said it has 11,000 paid reservations for its flagship luxury electric sedan, the Lucid Air.

Polestar said it plans to launch in nine more markets this year, doubling its global presence as it seeks to sell more of its electric sedans. The company, which is the electric performance vehicle brand under Volvo Car Group, also wants to double the number of retail stores to 100 locations and add more service centers by the end of the year. The Swedish automaker has more than 650 so-called “service points” in Polestar markets and wants to exceed 780 by the end of 2021.

REE Automotive has picked Austin for its U.S. headquarters. The company said the headquarters will help it address the growing U.S. market demand for mission-specific EVs from delivery and logistics companies, Mobility-as-a-Service and new technology players.

Tesla reported its second-quarter earnings and it was packed with news, including that the company generated $1.14 billion in net income, marking the first time the company’s quarterly profit (on a GAAP basis) has passed the three-comma threshold. And they hit that profitability metric without completely relying on the sale of zero-emissions credits to other automakers.

Tesla CEO Elon Musk weighed in on the company’s battery strategy and disclosed that the company is pushing the launch of its electric Semi truck program to 2022 due to supply chain challenges and the limited availability of battery cells. And everything is pointing to the Cybertruck also being delayed until next year.

And finally, Tesla’s latest quarterly earnings report showed growth in its energy storage and solar business. The company reported $801 million in revenue from its energy generation and storage business — which includes three main products: solar, its Powerwall storage device for homes and businesses, and its utility storage unit Megapack. More importantly, the cost of revenue for its solar and energy storage business was $781 million, meaning that for the first time the total cost of producing and distributing these energy storage products was lower than the revenue it generated. That’s good news.

eVTOLs and other flying things

Joby Aviation completed the longest test flight of an eVTOL to date: Its unnamed full-sized prototype aircraft concluded a trip of over 150 miles on a single charge. The test was completed at Joby’s Electric Flight Base in Big Sur, California, earlier this month. It’s the latest in a succession of secretive tests the company’s been conducting, all part of its goal to achieve certification with the Federal Aviation Administration and start commercial operations.

Lilium, the electric air taxi startup, has tapped German manufacturer Customcells to supply batteries for its flagship seven-seater Lilium Jet.

People stuff

AEye, a lidar company, has been adding to its executive team in the past few months. The most recent is the hiring of automotive veteran and former Valeo executive Bernd Reichert as senior vice president of ADAS. the company has also hired Velodyne’s former COO Rick Tewell, Bob Brown from Cepton and Hod Finkelstein as chief research and design officer from Sense Photonics.

Cruise is also on a bit of an executive and engineering hiring spree. The company sent me a list of recent folks who have joined including former Southwest Airlines employee Anthony Gregory as VP of market development, Phil Maher, the former Virgin Atlantic COO, as VP of central operations and Bhavini Soneji as VP of product engineering. Soneji was most recently VP of engineering at Headspace, and was at Microsoft and Snapchat before that.

Cruise also hired Vinoj Kumar, who oversaw Google’s cloud infrastructure and software systems, as VP of Infrastructure and Yuning Chai, former lead perception researcher at Waymo, as head of AI Research. In all, Cruise now employs more than 1,900 people.

Don Burnette, the co-founder and CEO of self-driving trucks company Kodiak Robotics, sat down with TechCrunch as part of our ongoing Q&A series with the founders of transportation startups. The interview covers a lot of ground, including Burnette’s views on the company’s strategy, current funding conditions in the industry and what he learned at Otto. the self-driving trucks startup he co-founded and that was acquired by Uber.

Trevor Milton, the fast-talking showman founder of Nikola and the electric truck startup’s former CEO and executive chairman, was charged with three counts of fraud. He is free on $100 million bail.

Milton “engaged in a fraudulent scheme to deceive retail investors” for his own personal benefit, according to the federal indictment unsealed by U.S. Attorney’s Office in Manhattan. Milton was charged with two counts of securities fraud and wire fraud by a federal grand jury.

 

They’re programmed to work hard and play hard

By Brian Heater

Industrial robotics are big and heavy — and in some cases, legitimately dangerous. They’re also extremely difficult to train — particularly if you plan to implement them for tasks outside of their purpose-built intentions.

There’s huge opportunity for the right AI/software company to come along and help make the bulky systems intended for things like auto manufacturing easier to program and more versatile. Honestly, there’s probably enough room to support multiple companies in the category as robots become an increasingly essential part of how we do business.

This week we saw a pair of big news stories from companies operating in that space. On Tuesday, Covariant announced an $80 million raise — a quick follow-up to the $40 million Series B it announced in May 2020.

Image Credits: Covariant

I spoke to president, chief scientist and co-founder (and recurring TC Sessions: Robotics guest) Pieter Abbeel for the piece, which you can check out here. I further picked the long-time UC Berkeley professor’s brain about some broader robotics trends.

We’ve seen a marked increase in investment activity around robotics and automation since the beginning of the pandemic. Do you anticipate that this interest will maintain?

It won’t just maintain. It’ll continue to accelerate on a dramatic scale. The demand isn’t new but the pandemic has certainly increased demand for resilient and robust robotics. COVID-19 accelerated a timeline that was already in motion. Other factors that contribute to the momentum include the rise of e-commerce replacing in-store purchases along with Amazon’s strive for efficiency. They’ve raised consumer expectations of fast delivery across the board and making good on that promise often starts with warehouse automation.

As someone with experience in both an educational setting and a startup, how have universities’ approach to incubating companies evolved. What more can and should be done to foster entrepreneurship?

With AI the transition from research to practice has been exceptionally fast. An idea could be published today, and many companies might be implementing it into their systems the next day. This trend has made AI researchers uniquely positioned to build new applications (compare this to, let’s say, Airbnb, Uber, food delivery companies, etc., which were not enabled by research advances, but by everyone having a smartphone, enabling a new model of doing business).

Structurally, one clear change at many universities is the introduction of artificial intelligence across many programs. A great example is “The Business of AI” course, which I co-teach in the Haas Business School at Berkeley, and which gives business students a solid understanding of the role of AI today, as well as trends and what the future might bring.

To foster more entrepreneurship in the U.S., leadership should consider how many international students are also the leading AI researchers. A faster visa/green card process for entrepreneurs would have a very high impact.

Do you foresee continuing to teach, as Covariant grown?

Yes. I see a very strong synergy between being at the forefront of academic AI research at Berkeley and being at the forefront of industrial R&D bringing AI Robotics into the real world as chief scientist at Covariant. The culture our CEO Peter Chen has fostered at Covariant also has great alignment with this; curiosity and lifelong learning are core values at Covariant.

How actively does your team consider biases in its AI work?

Bias in AI systems is of course a broader industry issue and is on the minds of our team members. As of today, bias in AI systems doesn’t directly play a role in our current robotic warehousing efforts. However, quality assurance more generally is core to everything we do, and quality assurance isn’t a one-axis thing, we have to consider quality and coverage of various data sources and performance across SKUs, warehouses, customers, etc. In that sense, there are actually many technical parallels.

It seems like most of the activity on the industrial robotics front is happening on the software/AI side. Are robotics manufacturers continuing to evolve their hardware as software improves?

Indeed, while we largely focus on the software/AI ourselves, we work with amazing partners to deliver fully functioning robotic systems. In doing so, we see continual improvement on the hardware as well. Most visible over a short time period are continual changes in end-of-arm tooling. In addition, we see interesting multiyear roadmap ideas in robotic arm form factors that take more R&D and design effort to bring to market.

Image Credits: Gramazio Kohler Research, ETH Zurich

The other big news of the week is the unveiling of Intrinsic, Alphabet’s most recent robotics play. Or, I guess I should say, most recently announced robotics play. The Alphabet X spinout has apparently been in the works for about five years now. It follows a fairly uneven robotics track record for Alphabet/Google that involved brief ownership of Boston Dynamics. But the company’s offering seems much more in-line with what Google excels at.

Here’s Intrinsic CEO, Wendy Tan-White, who most recently served as Alphabet’s VP of Moonshots:

Over the last few years, our team has been exploring how to give industrial robots the ability to sense, learn and automatically make adjustments as they’re completing tasks, so they work in a wider range of settings and applications. Working in collaboration with teams across Alphabet, and with our partners in real-world manufacturing settings, we’ve been testing software that uses techniques like automated perception, deep learning, reinforcement learning, motion planning, simulation and force control.

Image Credits: Agility

Closing the week’s roundup with a pair of athletic ‘bots. First is the return of Cassie, Oregon State University’s bipedal robot. Cassie took a bit of a backseat to OSU spinoff Agility’s delivery robot, Digit, but the school is continuing to do interesting things with the platform. A team of research helped teach the robot to run, using a a deep reinforcement learning algorithm.

In fact, Cassie managed to run a 5K in 53 minutes. Not great by human standards, but extremely solid for a robot using a single battery, particularly when you factor in the 6.5 minutes of troubleshooting an overheated computer and a poorly maneuvered turn.

Outside Olympians and T-shirt vendors, Toyota may well have been the most disappointed about the initial decision to delay the summer Olympics. The automotive giant clearly envisioned the Tokyo games as an ideal opportunity to showcase its technology for the world.

Now that the games are on, the company’s basketball robot CUE is back in a big way. After debuting in 2018, CUE returned to sink three-pointers during half-time at the USA-France game.

❌