GaP: A Graph-as-Policy Multi-Agent Self-Learning Harness For Variational Automation Tasks

2026-07-06

AI Summary

This work addresses the limited reliability of model-free policies in variational automation tasks, where drastic changes in object geometry and pose pose significant challenges. To this end, the authors propose the Graph-as-Policy (GaP) framework, which uniquely integrates interpretable graph-structured policies with a multi-agent self-learning mechanism. GaP constructs directed computational graphs comprising perception, planning, and control nodes, and iteratively optimizes their structure within a parallel simulation environment built upon MORSL, TAMP, and ROS. Evaluated on eight newly released variational automation benchmarks—four in simulation and four on physical systems—GaP substantially outperforms existing baselines, achieving markedly higher task success rates and throughput.

Abstract

For robots to work reliably in commercial and industrial applications, can recent advances in agentic coding systems combine interpretable robot programming with the open-world adaptability of model-free policies? We focus on "Variational Automation" (VA), a class of tasks that have larger variations in object geometry and pose than fixed automation. Model-free policies often struggle to close the reliability gap for VA tasks, which must be executed persistently and reliably in commercial and industrial applications. Motivated by prior work on Task and Motion Planning (TAMP) and the Robot Operating System (ROS), we introduce Graph-as-Policy (GaP), a multi-agent coding harness that generates directed computation graphs with perception, planning, and control nodes from a Modular Open Robot Skill Library (MORSL). GaP then 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 baselines. Details, code, and data can be found online: https://graph-robots.github.io/gap

Problem

• Variational Automation
• reliability gap
• model-free policies
• object geometry variation
• industrial robotics

Innovation

• Graph-as-Policy
• Variational Automation
• Multi-Agent Self-Learning
• Task and Motion Planning
• Modular Robot Skills