The August 2022 issue of IEEE Spectrum is here!

Close bar

Toyota AI Ventures Wants to Fund Your Mobile Manipulation Startup

Early-stage robotics startups can get up to $2 million in funding from Toyota to work on home assistive robots

5 min read
Toyota HSR Partner Robot
Toyota's HSR mobile manipulation robot.
Photo: Toyota

Today, Toyota AI Ventures, in partnership with the Toyota Research Institute, (TRI) is opening a “call for innovation” with the goal of throwing money at anyone with good ideas for “improving mobile manipulation technology for assistive robots that can help people in and around the home.”

Launched last year with US $100 million to spread around, Toyota AI Ventures has already started funding robotics companies, including Intuition Robotics and Realtime Robotics. While they seem to have not all that much trouble finding companies to give money to, this new approach is a bit more aggressive, perhaps with the aim of encouraging entrepreneurs who aren’t sure whether or not they can get funding for a robotics startup to actually give it a try. Here’s the info from the press release: 

Designed to spur entrepreneurial innovation by identifying key technology gaps, the initiative uses a call-and-response approach to offer promising startups the opportunity to secure from $500,000 to $2 million in venture capital funding from Toyota AI Ventures, as well as the possibility of partnering on a proof of concept project with TRI. 

The first call, developed in conjunction with TRI’s world-class robotics group, focuses on improving mobile manipulation technology for assistive robots that can help people in and around the home. Future calls may address technology challenges in other areas that TRI is working on as part of its research in robotics and automated driving, such as perception, machine learning, or simulation.

This first call for innovation is open to hardware and software startups around the world that have: (1) raised less than $3 million in funding; (2) can demonstrate their solution via a working prototype; and (3) have a strong business model to deliver value to customers. Examples of mobile manipulation solutions in hardware include safe, lightweight arms; grippers designed for common daily tasks; and technologies for better tactile sensing. Software solutions could include ways to compensate for lower-precision, lower-fidelity hardware; algorithms to learn from or annotate data; and ways to apply lessons learned from simulation. 

Toyota AI Ventures is accepting applications for this call for innovation now through the end of October, and submissions will be reviewed on a rolling basis. Qualified startups will be evaluated on the basis of their team, technology, business model, and go-to-market strategy.

Innovators are encouraged to learn more and apply on the Toyota AI Ventures website.

Mobile manipulation is a particularly tricky area for robotics right now, so we sent the folks at Toyota some questions about why they picked this and what kinds of things they’re hoping to see.

IEEE Spectrum: This first call for “improving mobile manipulation technology for assistive robots that can help people in and around the home” is very specific. Why choose to start here as opposed to other areas in robotics?

Jim Adler, managing director of Toyota AI Ventures: We recognize the challenge, both from a technically and commercially viable standpoint, of assistive robots to help people in and around the home, which is why funding opportunities in this area have been limited.  Our goal with this specific call was to challenge entrepreneurs to be creative in finding ways to enter this market.  We understand that it is not the “safe” bet, and are willing to take the high risk on very high impact applications, and want to encourage startups to venture into this uncharted territory.

What kinds of tasks in and around the home are a realistic opportunity for robotics right now? What tasks are probably too difficult for robots to solve in the near future?

Max Bajracharya, director of robotics at TRI: While we don’t want to constrain entrepreneurs who may have out-of-the-box ideas, three broad categories we think about are mobility, social assistance, and manipulation. Mobility could include moving items around for people, or moving people themselves. Social assistance could include connecting people with others remotely, finding new ways to motivate them, or socially or emotionally assist them. Manipulation could include picking up and putting away items, or helping people perform tasks. With certain constraints, there could be feasible applications in all of these categories. Trying to achieve a truly general-purpose robot that does all of these is likely too difficult for a startup to achieve in the short term.

What are some specific problems that you think need to be solved before mobile manipulation technology can find a useful and efficient place in the home?

Max Bajracharya: There are two classes of problems that could be addressed. The first is taking existing technology, finding a particularly important or valuable application of it, and then incrementally improving the technology to achieve this goal. For example, say an autonomous utility cart that moves stuff around from point to point at a certain price point, or last-mile delivery; while challenging, existing sensors, hardware, and technology are likely close to ready to do this, but it needs significant effort to make it robust and reliable.

The second is fundamental problems, for which breakthrough solutions could enable key or many applications (for example, robust grasping of objects among clutter, or very low cost but accurate torque sensing on joints, or software and hardware to make use of high fidelity tactile sensing for identifying objects in-hand). Both of these classes ultimately require a robot to be affordable, reliable, robust, and safe when performing the task it is intended to do, which are some of the key barriers to making robots useful and efficient in the home.

“The ability to transfer capabilities learned in simulation to a real system could make robots useful enough to start driving down the cost of existing hardware. Alternatively, a sufficiently low-cost but highly capable manipulator could enable many people throughout the world to develop software for it”

How much of the innovation in this space do you expect will come from software as opposed to hardware?

Max Bajracharya: Ultimately, breakthroughs in both hardware and software will likely be necessary to achieve safe, affordable, assistive robots for homes.  However, in the short term, innovations in software, hardware, or a combination could be the key catalyst to achieving this goal. For example, the ability to transfer capabilities learned in simulation to a real system could make robots useful enough to start driving down the cost of existing hardware. Alternatively, a sufficiently low-cost but highly capable manipulator could enable many people throughout the world to develop software for it; or clever ways to use software to compensate for existing but low fidelity hardware could enable it to be cost effective.
You ask for startups that are able to “demonstrate their solution via a working prototype.” I’m curious about how you evaluate demos and prototypes like these, since it’s often difficult to go from a robotic technology that works in a demo to a robotic technology that can be cost effectively and reliably deployed in the semistructured to unstructured environments that you find in homes.

Jim Adler: We have assembled an internal panel of experienced roboticists, business strategists, and investors to evaluate startups that apply. A key advantage of Toyota AI Ventures is that it has direct access to TRI’s staff of engineers, with extensive experience in the robotics field, and widely varying expertise (from deploying self-driving cars on public roads, to landing rovers on Mars, to developing and shipping consumer products, to funding and developing highly innovative breakthrough technology, to professors at top institutions).

We expect at least several members of the evaluating committee to have deep expertise in the technology being demonstrated, with a thorough understanding of how much effort is required to take a prototype to production, and to achieve the reliability required for unstructured environments.

[ Toyota AI Ventures ]

The Conversation (0)

Inside DARPA’s Subterranean Challenge

What SubT means for the future of autonomous robots

14 min read
A photo of a robot lighting up a stone tunnel.

An ANYmal robot from Team Cerberus autonomously explores a cave on DARPA’s Subterranean Challenge course.

Evan Ackerman

Deep below the Louisville, Ky., zoo lies a network of enormous caverns carved out of limestone. The caverns are dark. They’re dusty. They’re humid. And during one week in September 2021, they were full of the most sophisticated robots in the world. The robots (along with their human teammates) were there to tackle a massive underground course designed by DARPA, the Defense Advanced Research Projects Agency, as the culmination of its three-year Subterranean Challenge.

The SubT was first announced in early 2018. DARPA designed the competition to advance practical robotics in extreme conditions, based around three distinct underground environments: human-made tunnels, the urban underground, and natural caves. To do well, the robots would have to work in teams to traverse and map completely unknown areas spanning kilometers, search out a variety of artifacts, and identify their locations with pinpoint accuracy under strict time constraints. To more closely mimic the scenarios in which first responders might utilize autonomous robots, robots experienced darkness, dust and smoke, and even DARPA-controlled rockfalls that occasionally blocked their progress.

Keep Reading ↓Show less