AgenticRoboRL post-trains robot planners across different harnesses
Original title: AgenticRoboRL: Bringing Coding-Agent Post-Training to Agentic Robotics
On LIBERO-PRO, a 9B planner rose from 18.1% to 71.0% success without external memory.
AgenticRoboRL post-trains robot planners with supervised fine-tuning and multi-turn reinforcement learning while keeping harness execution fixed. Harness-specific adapters link planner interactions to task outcomes for VLA skill invocation and control-code generation. The authors say they will release the framework and trained planner checkpoints.
- Success reached 71.0%, near Codex with external memory at 72.1%.
- Adapters support VLA skill invocation and control-code generation.
- The common RL pipeline leaves execution components unchanged.
From the original article
1 min read at huggingface.co
Agentic robotics systems combine planners with robot harnesses to perform complex manipulation tasks. Recent work improves these systems through better harnesses and external memory, but relies on powerful closed-source models as planners and does not improve the planners' own capabilities. In contrast, coding agents have improved substantially by post-training the model itself.
By Xinxin Liu, Bingxin Xu, Wenrui Bao · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.