Jay Lee doesn’t recommend getting sued by Tesla if you’re trying to get your startup off the ground. But he thinks his company, Procession, may be better off for enduring the experience.
“I think this is like a resilience test or a pressure test,” he told TechCrunch in an exclusive interview. “People say that what doesn’t kill you makes you stronger, right?”
Mr. Lee, who was the technical director of Tesla’s humanoid robot program Optimus, was accused last year by his former employer of absconding with trade secrets to launch Proception. But after months of trading legal blows, a settlement was finally reached with Tesla, which dismissed the lawsuit earlier this month. (Tesla did not respond to a request for comment.)
Now, Lee is free to tackle what he considers an even more difficult problem: getting robot hands to function like humans.
To help with this, Proception announced Monday that it has raised an $11 million seed round led by First Round Capital, with contributions from Y Combinator and early-stage fund BoxGroup.
Proception also announced Monday that it has shipped the first batch of its “highly dexterous robotic hand” to “researchers and robotics companies,” and is accepting orders for a broader range of orders. The goal, Lee said, is to become a top hand supplier to other companies that don’t want to spend the time or resources developing what’s known in the industry as “dexterous operations.”
Despite the avalanche of money and attention pouring into the world of robotics, Lee doesn’t think it’s enough to create a robot hand that truly mimics the human hand.
One of the most vocal people on this issue is, in fact, his old boss, Tesla CEO Elon Musk, who has said that robotic hands are one of the biggest engineering problems yet to be solved.
Although Mr. Musk has claimed that Optimus robots could begin working in factories within a few years, the consensus view is that robot hands are still years away from being on par with human hands. Kevin Lynch, director of Northwestern University’s Center for Robotics and Biological Systems, told the Wall Street Journal last year that his team believes it will be 10 years before robots are “functional and useful and capable of doing some of the things that humans do.”
Li believes Proception can do it much faster, mainly because of the way it collects data.
Currently, most companies training humanoid robots use teleoperators to train their systems. A human wearing a virtual reality headset can see what the robot sees and manipulate what’s in front of it, and the robot can learn from the commands you give it.
Li said a major drawback to this approach is that the remote operator cannot receive feedback from the object the robot is touching. This approach also limits the number of robots a company can have at any given time, Lee said.
Proception’s solution is a glove equipped with a sensor. By having human testers wear gloves (and headsets), Proception and its customers can collect “human hand interaction data without the need for a robot in the loop,” according to a Proception press release.
This same glove will also be worn on Proception’s developing hand, acting as a “skin” packed with sensors. According to Proception, the hand has 22 degrees of freedom and multiple joints in each finger, allowing for a “wide range of dexterous movements.”
Lee said this approach will also allow Proception and its customers to collect more detailed, task-specific data that can more accurately bring the robot hand closer to a human’s. He also believes the company is well-suited to scale.
“You need both hardware and data, and they need to work together to make[dexterity]work. Many companies focus only on hardware or prefer hardware and data (collection) without scalability,” he said. “We’re working on this very dexterous hardware and highly scalable data, and we believe this is the key combination to solving this problem.”
Bill Trenchard, the first round partner who led the investment in Proception, said this was a big reason to support Lee.
“I think they have the best hands on the market, probably the most sophisticated hands today, and the underlying data and models to back it up,” he told TechCrunch. “Dexterous manipulation is a very, very, very important part of the whole story of future humanoids, and as many people are saying, it’s kind of the last mile in getting the robot to true performance.”
Mr Trenchard also praised Mr Lee’s ability to remain calm in the face of a lawsuit from his former employer.
“He was very transparent with us when this became clear. I think the team did a great job of staying calm,” Trenchard said. “Jay is a very strong leader.”
Lee is also confident. After facing Tesla’s “hardcore litigation department,” he told TechCrunch that he wouldn’t be surprised if the company asked for help as Proception grows.
“I think it will,” he said.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
