Event
RoboDoctor: A Training-Free Framework for Self-Improving Robot Policies
1 reports · 1 sources
Heyang Yu and other authors propose RoboDoctor, a training-free recursive self-improvement framework: under fixed model weights, it enables general robot policies and language model agents to adapt during execution through three revision objectives: diagnosis, action execution, and persistent experience recording. On 54 RoboDojo-Sim task entries, the framework achieved a five-dimensional average score of 57.10 and a native success rate of 49.44%; the recorded interactions demonstrate policy switching, selective correction, and grasp correction. The authors state that these results characterize execution-time adaptation within the evaluated systems, and that controlled experiments are still needed to isolate the contribution of cross-episode experience.
Report timeline
- RoboDoctor: A Training-Free Recursive Self-Improvement Framework for Robot Policies
RoboDoctor is a training-free framework that, under fixed model weights, lets generalist robot policies and language-model agents adapt during execution through three revision targets: diagnosis, action execution, and persistent experience records. On 54 RoboDojo-Sim task entries, the framework achieved a five-dimension mean score of 57.10 and a native success rate of 49.44%; recorded interactions show policy switching, selective correction, and grasp revision. The authors note that these results characterize execution-time adaptation in the evaluated system and that controlled experiments are still needed to isolate the contribution of cross-episode experience.
arXiv cs.RO ·