Robot arm grasps cube 71 minutes in: model writes its own kinematics, no training
Scobleizer · x · 2026-09-22
- A repost resurfaces an overlooked demo: an LLM (Claude Fable 5.1) drives an ARX R5 robot arm with a wrist-mounted Intel RealSense D405 depth camera — no demos, no training, no VLA, no hand-eye calibration, no teleop; a fixed camera only records.
- The model read the URDF and wrote forward/inverse kinematics, a control server (interpolation, workspace envelope, IK pre-check, automatic fallback to gravity compensation), depth-based table-plane fitting and object detection, plus a visual servo loop: move ≤2 cm, look, correct — lateral error from 10 cm down to millimeters before grasping.
- Cube-to-tray took just 71 minutes from the prompt "see if you can do this".
More from AGI Musings
- Prediction: hospitals will keep their own tested open-model formularies in two years — MaziyarPanahi · 2026-11-12
- Packy revives Qwest CEO's 1998 'no bandwidth glut' quote for the AI oversupply debate — prestonpesek · 2026-11-12
- AI agent web traffic grew 8x in two months, bots now 54% of web traffic — omooretweets · 2026-11-12
- François Fleuret: superintelligent AI will deduce more and experiment far less — francoisfleuret · 2026-11-12
- a16z maps the B2A economy: AI agents are now most of web traffic and soon its buyers — venturetwins · 2026-11-12
- Jason: Anthropic is rebuilding every SaaS product into Claude—free for life — smtabatabaie · 2026-11-12