More stories

  • in

    How MIT's robot Cheetah got its speed

    There’s a new version of a very quick quadrupedal robot from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL). While four-legged robots have garnered no end of attention over the last couple years, one surprisingly quotidian skill has been elusive for them: running.That’s because running in a real-world environment is phenomenally complex. The quick pace leaves scant room for robots to encounter, recover from, and adapt to challenges (e.g., slippery surfaces, physical obstacles, or uneven terrain). What’s more, the stresses of running push hardware to its torque and stress limits. MIT CSAIL PhD student Gabriel Margolis and Institute of AI and Fundamental Interactions (IAIFI) postdoc fellow Ge Yang recently told MIT News: 

    In such conditions, the robot dynamics are hard to analytically model. The robot needs to respond quickly to changes in the environment, such as the moment it encounters ice while running on grass. If the robot is walking, it is moving slowly and the presence of snow is not typically an issue. Imagine if you were walking slowly, but carefully: you can traverse almost any terrain. Today’s robots face an analogous problem. The problem is that moving on all terrains as if you were walking on ice is very inefficient, but is common among today’s robots. Humans run fast on grass and slow down on ice – we adapt. Giving robots a similar capability to adapt requires quick identification of terrain changes and quickly adapting to prevent the robot from falling over. In summary, because it’s impractical to build analytical (human designed) models of all possible terrains in advance, and the robot’s dynamics become more complex at high-velocities, high-speed running is more challenging than walking. What separates the latest MIT Mini Cheetah is how it copes. Previously, the MIT Cheetah 3 and Mini Cheetah used agile running controllers that were designed by human engineers who analyzed the physics of locomotion, formulated deficient abstractions, and implemented a specialized hierarchy of controllers to make the robot balance and run. That’s the same way Boston Dynamics’ Spot robot operates.

    [embedded content]

    This new system relies on an experience model to learn in real time. In fact, by training its simple neural network in a simulator, the MIT robot can acquire 100 days’ worth of experience on diverse terrains in just three hours. “We developed an approach by which the robot’s behavior improves from simulated experience, and our approach critically also enables successful deployment of those learned behaviors in the real-world,” explain Margolis and Yang. “The intuition behind why the robot’s running skills work well in the real world is: Of all the environments it sees in this simulator, some will teach the robot skills that are useful in the real world. When operating in the real world, our controller identifies and executes the relevant skills in real-time,” they added.Of course, like any good academic research endeavor, the Mini Cheetah is more proof of concept and development than an end product, and the point here is how efficiently a robot can be made to cope with the real world. Margolis and Yang point out that paradigms of robotics development and deployment that require human oversight and input for efficient operation are not scalable. Put simply, manual programming is labor intensive, and we’re reaching a point where simulations and neural networks can do an astoundingly faster job. The hardware and sensors of the previous decades are now beginning to live up to their full potential, and that heralds a new day when robots will walk among us.In fact, they might even run. More

  • in

    Drone delivery nearer to take-off following latest FAA recommendations

    By Stanisic Vladimir — Shutterstock
    No seriously, drone delivery is coming. If you were skeptical before, an FAA committee just took a huge step with the compliance groundwork to make that a reality.The so-called Beyond Visual Line of Sight Aviation Rulemaking Committee (BVLOS ARC) of the FAA published its final report last week. The committee is charged with paving the way toward broader commercial use of drones in the U.S., and its findings are being widely applauded by many in the sector who have sought a broader scope for commercial drone operations, including in applications like search and rescue and delivery.

    “Around the world, commercial drones are saving lives, making jobs more efficient, inspecting infrastructure at scale, and growing the economy,” said Lisa Ellman, Executive Director of the Commercial Drone Alliance, an industry trade group. “But here in the U.S., existing regulations hold back the drone industry by unnecessarily applying incongruous standards and approaches designed for crewed aircraft. This ARC report outlines a common-sense, risk-based, performance-based approach that balances safety with innovation, and will enable drone-based operations to scale in the U.S. for the benefit of all Americans.”Also see: Watch these autonomous drones zip through the woodsIndustry advocates have argued that unlocking the BVLOS marketplace will advance progress across a number of areas, including sustainable transportation, carbon emission reduction, equitable access to medicines and vaccines, safer and more effective critical infrastructure inspection, emergency response, aerospace jobs, and domestic manufacturing.The chorus on the other end of the spectrum hasn’t been all that loud, perhaps a function of the relatively obscure rulemaking processes at work, to which the industry is paying close attention but average consumers may not be.A common industry argument is that the U.S. has lagged behind Europe in efforts to integrate drones into the National Airspace in large part due to the limitations of the regulatory framework and the federal bureaucracy’s struggle to move nimbly. The recent FAA report gives the clearest indications yet of what a coming BVLOS regulatory framework will look like. The committee gave recommendations on things like pilot training requirements, right of way, and rules for third-party providers, such as commercial delivery vendors. Groups like the Commercial Drone Alliance, a non-profit organization led by leaders in the commercial drone and advanced air mobility industries, have long advocated for such recommendations, an interesting case where industry leaders have felt hamstrung by a lack of government guidance.In January, Congress issued a directive to the FAA to finalize and disclose its BVLOS plans within 90 days, prioritizing rulemaking around the issue. More

  • in

    Giant 180-ton robot trucks are mining gold

    A mining outfit in Australia is making a big bet on big robots. Following a recent proof of concept at a gold mine, mining contractor MACA will retrofit a fleet of 100 very large vehicles to create one of the largest autonomous heavy equipment fleets in the world.This is a pretty significant rollout and a proverbial canary in the gold mine for the sector’s broader automation ambitions. With the world hungrier than ever for precious and rare earth metals, technology is increasingly called on to make mining operations more efficient and cost-effective while unlocking increasingly scarce resources.

    Powering the new rollout is autonomous heavy equipment company SafeAI and its Australian partner, Position Partners. This new generation of autonomous heavy vehicle technology is a major upgrade from the first generation retrofits, which had limited onboard processing power and took a long time to see ROI in most cases. Early versions of autonomous vehicle technology in the sector also operated with closed legacy systems, preventing mixed fleets from communicating. Industries like mining have had this tech for 20 years now, but the lack of accessibility means it hasn’t really taken off yet.Autonomy 2.0 is changing that. AI-powered and armed with multimodal sensors (lidar, radar, camera), these new systems have significant onboard processing power to reduce network reliance and enable fast decisions. It’s also open, interoperable, and vehicle-agnostic — meaning tech like SafeAI’s retrofit autonomy can be applied to any vehicle, at pretty much any age from any manufacturer.”This technology is a game changer for our business, our customers and our industry,” explains Shane Clark, MACA’s General Manager of Estimating and Technical Services. “SafeAI’s versatile, scalable solution is unmatched in our industry right now, and has profound implications for site safety, efficiency and cost-effectiveness. We expect to see quick takeup from our customers as they begin to see the tremendous impact of this technology.” This means much greater scalability — like this 100 truck agreement — to accelerate the rollout of autonomous equipment for industries that are ready. One big benefit of autonomy is that it creates far safer working conditions for on-site workers.Of course, it also means that extractive industries like mining are becoming increasingly efficient. As personal computing and battery technologies increasingly drive demand for mined resources, automation technologies are propelling those industries to ever greater capabilities, a cycle that warrants increasing vigilance. More

  • in

    Digital radar on a chip speeds autonomous vehicle adoption

    Uhnder radar-on-a-chip
    A new radar-on-a-chip is on the way this year, capable of mass production and 4D digital imaging, heralding a new chapter for autonomous vehicles and next-gen advanced driver assistance systems (ADAS). Uhnder, a firm based in Austin, Texas, expects its digital radar to be automotive qualified in April 2022 and will debut on consumer production vehicles later this year.Radar was late to the game in autonomous vehicles, early developers of which preferred lidar and visual camera sensing suites. But radar has distinct advantages. For one, it’s a fantastic sensor for collision avoidance, particularly when used in concert with other sensing modalities. One disadvantage, however, has been its weight and form factor.Uhnder believes it has solved this with its radar-on-a-chip solution, a compelling example of how ADAS and autonomous vehicle development is helping drive evolution in sensing. “Uhnder’s 4D digital imaging radar-on-chip is a next-generation product that demonstrates new ways to advance automotive safety to save lives,” said Douglas Campbell, president of the Automotive Safety Council. “Fatalities of vulnerable road users are now 20 percent of all roadway deaths in the US and even more in developing countries. ADAS technologies, such as pedestrian automatic emergency braking (P-AEB) that can reliably operate at night, can help reduce pedestrian fatalities per the latest report from the Insurance Institute for Highway Safety. Improved high-resolution perception sensors, such as Uhnder’s radar-on-chip can potentially help reduce this rising fatality category.”Other firms have pursued radar-on-a-chip development, including Imec, one of the big players in commercial radar. In 2020 Imec presented a chip that processes radar signals using a recurrent spiking neural network. Its compact size made it easily deployable in drones and autonomous mobile robots, applications where size and weight are determinant factors for sensing systems.Uhnder is taking square aim at the vehicle space and claims its technology delivers the industry’s first digital radar solution with better accuracy and the power to sense moving or standing objects, large or small, at both short and long distances in all weather and lighting conditions, all while mitigating mutual interference between other radars, a big concern with analog radar. “Digital radar provides 16 times better resolution, 24 times more power on target, and 30 times better contrast than today’s analog offerings, improving detection capabilities for better road safety for all users – drivers, passengers, cyclists, and pedestrians,” said Manju Hegde, CEO and cofounder of Uhnder, Inc. “As more and more radars are fitted onto vehicles and other mobility solutions, interference among adjacent radar becomes problematic. Our radar, based on Digital Code Modulation, mitigates this problem.”The first tier 1 automotive customer for the 4D digital radar chip will be Magna, and Uhnder hopes to expand its customer base quickly.”I am excited with the progress made over the years and ready for the entire ecosystem to experience what we already know to be true – that digital radar is fundamental for next-generation ADAS,” said Swamy Kotagiri, CEO, Magna International. More

  • in

    Robot fry cook gets job at 100 White Castle locations

    White Castle
    White Castle seems to be all-in on its latest employee, a robotic fry cook. Flippy 2, the fast-food robot by Miso Robotics, will now be whipping up burgers and other food in 100 standalone locations.

    The news is part of a larger shift underway in the quick-serve sector, driven in part by the demand for contactless service and in part by a tight labor market and rising wages toward automation. Just this week, in a similar move, Jamba announced it was strengthening its collaboration with Blendid, which makes a juice robot.White Castle first trialed Miso’s original Flippy robot in a Chicago-area location in 2020. The burger chain, which bills itself as the first hamburger fast-food chain in-country (it was founded in 1921), then rolled out a version of Flippy, Robot-on-a-Rail (ROAR), to an additional 10 kitchens.”Artificial intelligence and automation have been an area White Castle has wanted to experiment with to optimize our operations and provide a better work environment for our team members,” said Lisa Ingram, CEO of White Castle, at the time. “We believe technology like Flippy ROAR can improve customer service and kitchen operation. This pilot is putting us on that path — and we couldn’t be more pleased to continue our work with Miso Robotics and pave the way for greater adoption of cutting-edge technology in the fast-food industry.”Also: Sam’s Club betting its cleaning robots can do double dutyThe sales pitch by Miso is that its robot can alleviate inefficiencies in the back of house while ensuring consistent quality. Given the scope of the rollout, White Castle clearly deems the ROI equation valid.”We could not be more grateful for the confidence White Castle has shown in us as we enter into the next phase of our partnership,” said Mike Bell, CEO of Miso Robotics. “White Castle was the first large brand to embrace our technology,  and we are thrilled that our Flippy pilot made such a positive impact on their operations that they want to integrate 100 more. We can’t wait to continue on this journey with such an outstanding partner.”         

    Miso’s journey, which we’ve covered since the company came out of stealth, has been fun to watch. The company did a non-traditional crowdfunding campaign and is primarily funded by individual investors. It boasts over 15,000 shareholders and a whopping  $50M in crowdfunding to date. Its E round gives it a market valuation of $500 million. More

  • in

    Smoothie bots to take over Jamba locations

    Robots are coming to the quick serve space, and it’s been fascinating to track the progression across various food offerings. In the latest iteration, smoothie purveyor Jamba is expanding its partnership with food automation firm Blendid on additional robotic smoothie kiosks.Last year I spoke with the co-founder of Blendid, Vipin Jain, about the moment contactless service is having since the onset of the pandemic and the business proposition a company like Blendid can offer potential partners.”From the operator’s perspective, the benefits of deploying autonomous robotic kiosks such as Blendid are all tightly tied to supporting their business,” said Jain. “Automated robotic food platforms provide cost-effective and efficient fresh food options, opening the door for 24/7 operation with very minimal downtime to quickly restock fresh ingredients. The ability to offer fresh food in areas where it may not make operational sense otherwise, due to a small footprint or low traffic, opens new revenue streams. In fact, we’ve provided businesses (operators) with a very attractive unit economics – with cash-on-cash return of less than 18 months … almost unheard of in the food service industry!”Jamba, which along with Walmart is one of Blendid’s early partners, certainly seems to be seeing the business case. The partnership is expanding to introduce new kiosks on college campuses, part of an interesting trend industry-wide of using colleges as automation testbeds.”Since the beginning, we have wanted to test the Jamba by Blendid kiosks on college campuses. We know there is increasing demand for more ways to access our products and by introducing our robotic kiosks at colleges and universities across the country, we are making it even easier for consumers to enjoy smoothies 24/7,” said Geoff Henry, president of Jamba. “It has been rewarding to see guests engage enthusiastically with our kiosks, and we are looking forward to speaking with future operators who are interested in testing with us and learning more about the opportunity.” The food robotics market, estimated at $1.9 billion in 2020, is expected to reach $4.0 billion by 2026. Advances in robotics and AI, coupled with operational cost advantages and major consumer and retailer shifts, are driving the food industry to more rapidly embrace automation. A tight labor market and well-publicized worker shortages among restaurant employers are also hastening the shift. “Automated food solutions were growing before the pandemic,” Jain told me, “but the COVID-19 crisis poured gasoline on the fire. Business continuity became a challenge for many food service and retail companies. Suddenly businesses had to figure out how to offer food safely in a contactless manner and cost effectively. The pandemic has pushed up the timeline considerably. Based on the exponentially growing interest Blendid is receiving from prospective operators worldwide, mounting staffing challenges, and robotics cost reductions, I expect food robots to be pervasive within 5 years. What used to be forward-thinking has become the current-thinking. This is the new “normal” for food service.”

    All to say, these shifts are the beginning of a much larger change afoot. Expect to see robots behind the counter sooner than expected. More

  • in

    How cell tower 'COW' drones will keep fans safe at the Super Bowl

    FirstNet flying COW
    For fans in Los Angeles, host to this year’s Big Game, the Super Bowl is a chance to tailgate and celebrate. For first responders in the LA region, it’s a massive public safety event that’s a year in the making.One of the big concerns is network strain. AT&T is answering the call for help with a number of technologies, including the “flying COW,” essentially drones that double as cell towers.Also: How to watch the Super Bowl: All of your streaming optionsThe problem is acute. LA’s SoFi stadium has 10X the capacity versus the Super Bowl’s venue in 2020, and there’s a long legacy of network disruptions during large events. That can lead to significant safety concerns. One striking example is the 2013 Boston Marathon bombing. So many bystanders were using their cell phones that the high call volume saturated the local cellular networks, causing signal delays and failed calls that rendered mobile phones nearly useless. For area first responders on duty that day, the consequences of poor network coverage and capacity was dire — the lack of a reliable network delayed the ability to share the images and videos that ultimately helped to identify the alleged perpetrators.

    [embedded content]

    AT&T has been busy mobilizing around SoFi Stadium. Thanks to the provider’s public-private-partnership with the First Responder Network Authority (FirstNet Authority) — an independent agency within the federal government — AT&T is also delivering 5G to first responders in Los Angeles. FirstNet has been committed to revamping public safety for massive events and also during disaster recovery. The three Flying COWs (which stands for “cellular on wings”) in the FirstNet fleet comprise two tethered drones and a trailer for transport that is equipped with a satellite dish and fiber connections. Southern California is a fair weather place, but the drones are capable of withstanding light rain and wind speeds up to 25 miles per hour — while reaching heights of up to 400 feet. Outfitted with Band 14, AT&T can use the COWs to help equip FirstNet subscribers in the US with connectivity during high-usage events. 

    For the Super Bowl, FirstNet will also deploy a Communications Vehicle, a therapy dog to help with first responders’ mental health during stressful situations, one FirstNet Micro SatCOLT (Satellite Cell on Light Truck), six ground COWs (Cell on Wheels) with Band 14 spectrum antennas for FirstNet, and thousands of feet of fiber in and around the stadium.AT&T has been moving aggressively in Los Angeles. From 2018 to 2020, it expanded coverage and improved connectivity by investing nearly $2.7 billion in wireless and wireline networks in the greater Los Angeles area. More

  • in

    With lead in China, RoboTaxi maker turns sights on U.S.

    AutoX expands RoboTaxi empire to San Francisco.
    If the race for autonomous vehicles is measured in absolute numbers, a company that’s been surprisingly successful navigating real-world rollouts in both China and the U.S. is winning. AutoX now counts more than 1,000 of Level 4 autonomous RoboTaxis in operation in China, and it’s been a surprise front runner in U.S. L4 testbeds as well.

    Electric Vehicles

    The thousand fleet milestone comes as AutoX is riding a wave of recent announcements. In July 2021, AutoX’s newest Gen5 system-equipped RoboTaxis started rolling off the production line. More recently, in January 2022, AutoX shared an inside look at its end-of-line, Level 4 fully driverless RoboTaxis dedicated production facility located near Shanghai, China, with a video. Since starting production, the RoboTaxi assembly line has been in full operation. Back 2020, capitalizing on a COVID-19 pandemic-induced emphasis on contactless services, the company made headlines with a dual-country approach to autonomy testing and market rollout, essentially cornering the RoboTaxi market in Shanghai while also winning a coveted permit to test its driverless cars without drivers in California, becoming just the third company to be awarded the permit. The rapid rise is all the more impressive, given that it seemed to get a late start in the driverless race. AutoX Founder and CEO Jianxiong Xiao, who was the founding director of Princeton’s Computer Vision and Robotics Labs before leaving the school in 2016 to found AutoX, started his company with modest seed funding after moving his family from Princeton, NJ, to Silicon Valley in 2016. His value proposition was that inexpensive cameras paired with the right AI would be enough for safe L4 autonomous driving. Unlike competitors, his company operated in stealth until very recently, although a California DMV filing to test self-driving vehicles put him on insiders’ radar early on. Behind the scenes, Jianxiong’s computer vision bona fides (he’s an all-star in the field) have helped him attract major academic talent.Using inexpensive sensors, Jianxiong says he is on a mission to democratize autonomy via cutting edge AI. The notion of democratizing autonomous driving is embedded deep in AutoX’s DNA. It’s a rare bootstrapped company in an ecosystem dominated by the likes of Uber, Intel, and Google. With ample presence in China, AutoX is redoubling its U.S. efforts. The company has launched a new RoboTaxi operations center in San Francisco, representing a unique road challenge for autonomous vehicles and in one of the densest urban centers in the U.S. “We want to deploy the AutoX RoboTaxi fleet in many cities to serve hundreds of millions of people, as well to improve and become a part of communities around the world. Cities such as San Francisco, Shenzhen, Shanghai, and Beijing are the launchpads for RoboTaxis to transform people’s daily lives. And that’s just the beginning,” says Jianxiong.  More