Proception Secures $11M to Master the Final Frontier of Robotics: Human-Like Dexterity
Proception, a startup focused on the complex challenge of robotic dexterity, has settled a trade secret lawsuit with Tesla and secured $11 million in funding. The company is pioneering new ways to train robot hands using specialized data collection methods.
One of the most persistent "grand challenges" in robotics is the replication of human hand dexterity. Proception, a robotics startup that recently emerged from a legal battle with Tesla, is tackling this problem head-on with a new $11 million investment. The company’s focus is on the intricate interplay of tactile feedback and motor control required for fine-motor tasks—something that traditional robotic grippers still struggle to master.
Proception’s approach involves a unique data collection pipeline that goes beyond simple visual imitation. By recording the micro-pressures and subtle adjustments humans make during tasks like handling delicate components or using hand tools, the company is building a high-fidelity dataset for training neural networks. This "haptic-first" training allows their robot hands to adapt to different textures and weights in real-time, moving closer to the versatility of a human laborer.
The settlement with Tesla, which centered on trade secrets related to humanoid development, clears the runway for Proception to scale its operations. Historically, robotics has succeeded in "pick-and-place" tasks in controlled environments. Proception’s goal is to bring that same reliability to "manipulation-heavy" industries like electronics assembly and warehouse fulfillment, where the ability to grasp and manipulate varied objects is the key to full automation.
Source: TechCrunch
