Coding agents find better pathfinding strategies for robots and warehouses
Policy Search in Code Space for Multi-Agent Pathfinding
RoboticsArtificial Intelligence
Summary
Finding good ways for many robots or agents to move around without bumping into each other is hard and usually requires lots of tricky hand-coded rules or black-box learning methods. The authors show that letting AI coding agents write and improve the robot control code directly, guided by tests and feedback, can discover better strategies. They tested this approach on robot racing maps and warehouse simulations, where it found new, provably safe and efficient methods that outperform human-made and some existing commercial ones. This means AI can help code complex coordination tasks with clear logic and strong performance.
What this means in practice
- •For warehouse automation teams: Improve robot coordination methods in complex warehouse layouts, enhancing throughput beyond current proprietary solutions.
- •For robotics software developers: Use AI-generated code to create and refine multi-agent pathfinding policies that safely scale to dense and asynchronous robot movements.
Authors
Matthias Eder, Pantelis Emmanouil, Vivek Farias, Fabian Fischer, Adam Jozefiak, Tianyi Peng, Thomas Ulz, Raphael Watschinger, Fangyu Wu, Andrew Zheng
Abstract
Developing coordination policies for complex multi-agent systems relies on either labor-intensive algorithm design or opaque task-specific reinforcement learning. With recent advances in autonomous coding agents, we show that policy search directly in code space offers a compelling third path that combines automated optimization with explicit interpretability. We investigate a human-directed workflow in which coding agents iteratively propose, implement, evaluate, diagnose, and eventually refine algorithms guided by simulation feedback and explicit correctness verification. We conducted two case studies and found that this workflow robustly outperformed any manually engineered solutions. On a challenging map in the 2024 League of Robot Runners competition, a two-day search discovered a policy that exceeds the stronger locally reproduced competition entry by 35.0% on the densest instance and generalizes to unseen map instances. In simulated real-world warehouses with irregular lanes, continuous robot geometry, and asynchronous motion, a week-long search discovered an entirely new class of multi-agent pathfinding algorithm that outperforms a calibrated proprietary solution in throughput by 58.9%. Under stated assumptions, we further prove that the new algorithm is collision-free and completes, in finite time, every accepted goal that is not withdrawn or replaced. Together, the studies show that policy search in code space powered by state-of-the-art coding agents offers a highly competitive path to building real engineering solutions that are both effective and analytically tractable.