Claude Code controls real robot arm with no training, achieves mm-level accuracy via self-calibration
mishig25 · x · 2026-08-28
LeRobot demonstrated Claude Code operating a real SO-ARM101 robot arm. The process required zero training—no policies, teleoperation, or demonstrations—as the agent measured the workspace and wrote motion commands itself.
Key Highlights:
- Novel Calibration: No checkerboard or camera intrinsics needed. The arm uses itself as a ruler. It detects 16 points the software draws in the camera view by resting the gripper on them, achieving 4.1 mm position accuracy and 3.0 mm placement accuracy.
- Performance: The best run so far stacked 12 bricks with all four color groups formed.
Anthropic's multi-agent system (MHS) is currently in research preview, with open source coming soon.
More from coding & agent
- Paper: AI coding agents leave behavioral fingerprints, identifiable at 97.2% accuracy — thisguyknowsai · 2026-08-28
- Developers recommend Cloudflare Computer for AI agent environments — threepointone · 2026-08-28
- GLM-5.3-Flash built a full Blender scene autonomously in 12 hours — smtabatabaie · 2026-08-28
- How do developers decide to adopt an MCP server? Which signals matter most — SnooPuppers6082 · 2026-08-28
- Why do agent coding runs stop at 30 minutes? The missing machine-checkable feedback loop — EntireBig7258 · 2026-08-28
- No mobile app? User feeds Conductor's API docs to Grok Bot to manage cloud agents — iannuttall · 2026-08-28