Robots ace blind precision assembly using only joint encoder signals, no cameras
kevin_zakka · x · 2026-10-19
New robotics research shows every robot already carries contact sensors: its own joints. Tiny encoder drift when touching objects—already used for blind walking in legged robots—is enough for precise assembly. Policies using only joint positions can insert pegs, mesh gears, and thread nuts with no camera, tactile sensor, or object poses, and the peg policy transfers zero-shot to the real world.
More from Embodied
- ETH Zurich teaches a robotic hand to walk on its fingers with reinforcement learning — MIT_CSAIL · 2026-10-20
- LeCun co-authored LeHami: 1830s symplectic physics fixes world model collapse, 2.8x more accurate with 1/8 compute — alex_verem · 2026-10-19
- MolmoAct2 lead Duan Jiafei joins NUS with multimillion-dollar robotics grant — DJiafei · 2026-10-19
- PixVerse R2 acts like a world model: WASD movement and text inputs reshape a live game world — hey_abusiddik · 2026-10-19
- Robotics writer admits he was wrong: Physical Intelligence just shipped a model API — binarybits · 2026-10-19
- Why testing robot models is even harder than testing LLMs — the RoboDojo catch — binarybits · 2026-10-19