HexaShell: VLM Coding Agents Drive Dual-Arm Robots, SOTA on RoboDojo
HexaFuture · hf · 2026-10-19
HexaShell explores how far a VLM coding agent can go controlling a robot instead of a computer. The harness is thin — it only turns the agent's commands into robot motion. The agent drives a dual-arm robot through one command line per arm, sees three RGB cameras with known calibration, and after each failed attempt reads the log, finds the cause, and writes small tool programs to fix it, keeping what it learns as code. Tools may only use what the robot observes, locating objects via triangulation. It achieves state-of-the-art results on RoboDojo but struggles with contact-heavy tasks.
More from Embodied
- Robot brain inference cost fell 84% in two months, from $2.36 to $0.38 per task — victor_explore · 2026-10-19
- NORINCO unveils Fuxi, an anti-riot humanoid robot that mirrors its operator's movements in real time — CyberRobooo · 2026-10-19
- Has GPT-6-Astra Achieved Embodied Intelligence? A 12-Benchmark Evaluation — sensenova · 2026-10-19
- Amazon FAR's StopWatch: Explicit Goal Grounding Brings Object-Goal Navigation Near-Expert Success — Amazon-FAR · 2026-10-19
- Robot can now recover its balance after a 0.1 N push — _Stocko_ · 2026-10-19
- Chinese blogger builds working electric hoverboard with 10 motors and real-time AI voice coaching — churchkey · 2026-10-19