GaP: A Graph-as-Policy Multi-Agent Self-Learning Harness For Variational Automation Tasks
GaP: A Graph-as-Policy Multi-Agent Self-Learning Harness For Variational Automation Tasks
This research presents GaP, a multi-agent coding harness designed for variational automation tasks that feature significant variations in object geometry and pose. By combining interpretable robot programming with model-free policies, GaP generates directed computation graphs incorporating perception, planning, and control nodes derived from a Modular Open Robot Skill Library (MORSL). The system utilizes an internal simulation environment to rehearse task instances in parallel, iteratively refining graph structures and parameters to enhance success rates and throughput. Evaluation across eight new open benchmarks—four in simulation and four in real-world settings—indicates that GaP achieves success rates significantly outperforming baseline methods. The project provides details, code, and data via its associated repository.
The paper introduces GaP, a multi-agent system that generates computation graphs for variational automation tasks using an internal simulation environment to refine graph structures and parameters before real-world execution.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
- code: Reported pending Open source S5
- weights: not_reported
- dataset: not_reported
- code: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
rom a Modular Open Robot Skill Library (MORSL). GaP generates an internal simulation environment to rehearse task instances with different graphs in parallel to iteratively refine the graph structure and parameters to improve success rates and throughput. Evaluation with 8 new open VA task benchmarks, 4 in simulation and 4 in real-world, suggests that GaP can achieve success rates that significantly outperform baseli
Open source S4
nes. Details, code, and data: https://graph-robots.github.io/gap
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org