Robot Manipulation with GPT-6-Astra: Body Knowledge, Experience Reuse, Emergent Skills, and Sim2Real Transfer

What happened

arXiv:2609.31770v1 Announce Type: new Abstract: General-purpose multimodal agents can write robot-control programs, but repeated exploration and model-mediated action selection can make execution slow. In nine paired comparisons (18 trials) at starts displaced by 10-100 cm, experience recorded at the original start reduces mean time by 58-63% relative to no experience, demonstrating generalization to the tested new starting positions.

During experience experiments, GPT-6-Astra spontaneously generates a short visual-feedback program. Researcher-refactored versions reduce mean local-task time by 29-31% in 27 simulation trials.