RPG framework for embodied agents
Reconstruct, Practice, Go Real: Guided Self-Improvement for Embodied Agents
RPG demonstrates how autonomous systems can improve through simulation-based practice without weight updates, enabling rapid skill acquisition. Developers working on robotics and embodied AI can implement this approach to reduce human effort in skill development. The framework's success in both simulation and physical trials shows practical applicability for real-world robotics applications.
The Reconstruct, Practice, Go Real (RPG) framework improves robot task success from 28.
6% to 95.0% after 15 practice rounds on 22 manipulation tasks.
RPG identifies capabilities in offline datasets and constructs practice tasks in simulation without updating model weights.
It uses execution feedback and simulator state to diagnose failures, develop new skills, and refine existing ones.
The framework outperformed baselines including ASPIRE (75.5%) and CaP-Agent0 with GPT-6 Astra Pro (60.
0%).
Reconstruct, Practice, Go Real: Guided Self-Improvement for Embodied Agents
Building reliable robot capabilities across diverse tasks requires substantial human effort to develop and maintain skills, design rewards, and integrate perception with control. We present Reconstruct, Practice, Go Real (RPG), a framework for autonomous improvement of robot execution systems without updating model weights. RPG identifies manipulation capabilities in an offline dataset and constructs related practice tasks in simulation. During practice, RPG uses execution feedback, privileged simulator state, and available dataset videos to diagnose failures. It develops new reusable symbolic skills, refines existing skills, and revises the system prompt based on these diagnoses. Cross-task evaluation tests individual candidate changes and merged revisions before they are retained for reuse. At test time, a multimodal LLM uses the resulting system prompt and skill library to coordinate perception and robot control. On held-out initializations of 22 manipulation tasks, RPG improves task success from 28.6% after the first practice round to 95.0% after 15 rounds, outperforming all evaluated baselines, including ASPIRE (75.5%) and CaP-Agent0 powered by GPT-6 Astra Pro (60.0%). After a common calibration and hardware-adaptation procedure, the frozen system succeeds in all 30 physical trials, with ten trials on each of three tasks. Project Website: https://rpg-robot.github.io/
This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include RPG, robots.
Source attribution: curated from arXiv cs.AI (@Yen-Jen Wang, Haozhe Jiang, Shuying Deng, Haoru Xue, Weirui Ye, Rocky Duan, Nika Haghtalab, S. Shankar Sastry, Pieter Abbeel, Haozhi Qi).