Graph attention network spots robot collisions quickly and flexibly
CollisionGAT: Controller-Agnostic One-Step Collision Screening for Multi-Agent Motion
RoboticsMachine LearningMultiagent Systems
Summary
Robots moving together need to avoid bumping into each other or obstacles. The authors created CollisionGAT, a smart system that looks at where robots are and where they want to go, then quickly estimates the chance of crashes for each robot. This score helps any robot controller decide if the move is safe or needs fixing. They tested it using real geometric checks to make sure the system accurately predicts collisions.
What this means in practice
- •For robotics engineers: Integrate CollisionGAT scores to improve safety checks in multi-robot movement steps before execution.
- •For warehouse automation teams: Optimize coordination among multiple robots navigating obstacle-filled warehouses using fast collision risk evaluation.
Authors
Alan Debbas, Edwin Meriaux, Gregory Dudek
Abstract
Before a team of robots moves, each proposed step must be checked for collisions with other robots and with obstacles. We present CollisionGAT, a graph-attention network that reads the current and proposed states of moving agents together with locally relevant stationary obstacles and returns one collision-risk score per moving agent. Any controller can use these scores to accept, repair, replan, or postpone a proposed step. We mount CollisionGAT on a continuous path-following controller and on GATeD, an obstacle-blind D* Lite planner that uses typed vetoes to update its planning graphs. Exact geometric checks supply the training labels and independently audit every executed step.