Robotics firm Proception, founded by former Tesla technical lead Jay Li, has announced a significant milestone following the resolution of a trade secret lawsuit initiated by his previous employer. The startup confirmed earlier this month that it had reached a settlement with Tesla, which subsequently dismissed the legal action. This development clears the path for Proception to focus on its core mission: developing highly dexterous robotic hands capable of mimicking human capabilities.
Alongside the settlement news, Proception revealed it has successfully secured an £8.5 million ($11 million) seed funding round. The investment was led by First Round Capital, with additional contributions from prominent early-stage investors Y Combinator and BoxGroup. This substantial capital injection is expected to accelerate the company's research and development efforts, particularly in a field widely acknowledged as one of the most challenging in robotics.
Proception has also begun shipping the initial batch of its advanced robotic hands to researchers and other robotics companies, with wider orders now being accepted. The company's strategy is to become a primary supplier of these sophisticated hands, enabling other organisations to integrate advanced manipulation capabilities into their robots without the extensive resources typically required for in-house development. This focus on 'dexterous manipulation' addresses a critical bottleneck in the broader robotics industry, an issue even acknowledged by figures like Tesla CEO Elon Musk.
A key differentiator in Proception's approach lies in its innovative method for collecting training data. Unlike many current methods that rely on teleoperators controlling robots in real-time, Proception utilises sensor-equipped gloves worn by human testers. These gloves capture detailed human hand interaction data directly, removing the need for a robot in the initial data collection loop. This system not only promises more nuanced and task-specific data but also offers a more scalable solution for training robotic systems, potentially accelerating the development timeline for human-like robot dexterity.
The robotic hand itself, which also incorporates this sensor-packed glove technology as its 'skin', features 22 degrees of freedom and multiple joints per finger. This intricate design is intended to facilitate a wide array of precise and complex motions, bringing the functionality closer to that of a human hand. Proception believes this combination of advanced hardware and scalable data collection is fundamental to overcoming the long-standing challenges in achieving true dexterous manipulation in robotics.