Stigmergy-Based Multi-Agent Motion Planning Using QR Codes

Abstract

This thesis takes a stigmergy-inspired approach to asynchronous multi-agent motion planning in communication-restricted environments. Building on sampling-based motion planning, it introduces a framework for encoding paths within QR codes that are physically placed in the environment. Feasible paths generated by a robot are post-processed through path pruning and ellipsoidal containers to reduce data size while preserving essential trajectory information. QR codes allow later robots to decode, reconstruct, and replan directly within the corresponding ellipsoid, supported by algorithms for occupancy grids, greedy ellipsoid expansion, and uniform ellipsoidal sampling.

Simulation results show that the proposed approach reduces computation time and memory requirements while maintaining feasible path quality, highlighting the potential of geometry-based stigmergy as a lightweight, communication-free mechanism for sharing trajectory information among autonomous agents.