The startup Proception has ended a legal dispute with Tesla after settling a lawsuit in which the U.S. company accused Jay Li, formerly a technical lead on the Optimus humanoid robot program, of misappropriating trade secrets to establish his new company. The court dismissed the lawsuit earlier this month, while Tesla did not respond to TechCrunch’s request for comment.
Alongside the end of the dispute, Proception announced Monday that it had raised $11 million in a seed funding round led by First Round Capital, with participation from Y Combinator and the early-stage investment firm BoxGroup. The company has also begun shipping the first batch of its highly dexterous robotic hand to researchers and robotics companies, and has opened the door to broader orders.
Focus on manual dexterity
Proception aims to become a supplier of robotic hands for companies that do not want to devote the time or resources needed to develop what the robotics industry calls “dexterous manipulation,” meaning a robot’s ability to grasp and handle objects with precise and varied movements.
Li believes the robotics industry has received substantial funding and attention, but that an insufficient share of this focus has gone toward developing robotic hands that closely replicate the human hand. This is considered one of the most difficult engineering challenges in humanoid robotics; there is broad industry consensus that achieving hands comparable to human hands still requires many years. A previous report quoted Kevin Lynch, director of Northwestern University’s Center for Robotics and Biosystems, as saying that his team estimates it will take a decade before these hands become practical, useful and capable of performing some of the things humans do.
A glove for collecting interaction data
Most companies currently training humanoid robots rely on remote operators. The human operator wears a virtual reality headset to see what the robot sees and control the objects in front of it, after which the system learns from the commands it receives.
But Li says this approach lacks sufficient feedback from the objects the robot touches, and that its ability to collect data is limited by the number of robots available at any one time. To address this, Proception developed a glove equipped with sensors, worn by human testers along with a virtual reality headset, allowing the company and its customers to capture data on human hand-object interaction without needing a robot to be part of the process.
The same glove is used on the robotic hand Proception is developing, serving as a kind of sensor-filled “skin.” The hand has 22 degrees of freedom, with multiple joints in each finger, enabling a broad range of precise movements. Li believes this approach will help collect more detailed data tied to specific tasks, potentially making robotic hand movements more similar to human hand movements, and considers it suitable for scaling.
Funding to support hardware and data
Li said success in dexterous manipulation requires combining hardware and data, explaining that some companies focus on hardware alone or pair it with data-collection methods that are difficult to scale. Proception is instead betting on highly dexterous hardware with scalable data.
Bill Trenchard, a partner at First Round who led the investment, said this combination of the robotic hand, data and models was a primary reason for backing the company. He added that dexterous manipulation represents an important part of the future of humanoid robotics, describing it as almost the final stage in getting these robots to perform in practice.
Trenchard also praised Li and his team for continuing to work throughout the lawsuit. Li, for his part, said he would not be surprised if Tesla sought Proception’s help in the future as the company grows, reiterating his belief that this could happen.