Robot assembles diverse parts using only joint sensing
Blind Precise Assembly: Proprioceptive Sim-to-Real Peg-in-Hole for Diverse Parts
Robotics
Summary
Robots usually rely on cameras or touch sensors to assemble small parts, but this research shows they can do precise assembly using only their internal joint sensors. The authors had robots learn tasks like inserting pegs and threading nuts by feeling the movements and forces in their joints, without any vision or external touch data. Their approach works with many different parts and transfers well from simulation to the real world. They also created a tool to quickly simulate many assembly tasks, helping train these joint-sensing robots efficiently.
What this means in practice
- •For industrial automation engineers: Enable industrial robots to assemble diverse parts precisely without requiring vision or external tactile sensors, improving robustness in cluttered or occluded environments.
- •For robotics simulation developers: Use the authors’ open-source part generator to create accurate and efficient simulations of assembly tasks for training control policies.
Authors
Antonia Bronars, Younghyo Park, Kushal Kedia, Pulkit Agrawal
Abstract
Every robot already carries high-resolution contact sensors: its own joints. Legged robots and humanoids exploit them to walk, run, and track whole-body motions without vision, but there the contacts are large and forceful. Is joint sensing also enough for precise, contact-rich assembly, where the robot must find a millimeter-scale opening, mesh gear teeth, or catch a thread through light contact? We study this question by training blind assembly policies that observe only a short history of joint positions, velocities, and their own actions, without visual, force, torque, or tactile sensing. The poses of both the fixture and the part are unknown to the policy, and the grasp is randomized substantially. The policy must therefore assemble the parts through contact, sensed exclusively through its own proprioception. Trained with reinforcement learning entirely in simulation over randomized parts, grasps, and contacts, such policies learn peg insertion, gear meshing, and nut threading. The peg insertion policy transfers zero-shot to the real world and inserts pegs of different shapes and sizes in 73% of trials. Controlled comparisons with matched visual policies in simulation show that blind policies generalize far better to unseen part shapes and sizes, though they can be slower. Training such policies needs thousands of different assemblies, each simulated with its clearance intact, so we also release an open-source generator that builds them procedurally from exact convex pieces for MuJoCo; it is more accurate and faster than convex decomposition or MuJoCo's signed-distance collision. Videos and interactive simulations are at https://younghyopark.me/blind-proprioceptive-assembly/.