What question did this study set out to answer?

To develop a scalable solution for multi-agent path finding that adapts to dynamic environments using curriculum learning.

May 6, 20260 citationsOpen Access

View Full Paper

CBS-Bootstrapped MAPPO: Scalable Multi-Agent Path Finding Under Dynamic Obstacles and Natural Language Control

KKKyaw Linn Khant

Key Points

To develop a scalable solution for multi-agent path finding that adapts to dynamic environments using curriculum learning.
Introduced CBS-Bootstrapped MAPPO, leveraging CBS as a decaying reward signal.
Implemented a three-phase curriculum for training multi-agent reinforcement learning (MARL) policies.
Utilized natural language commands for zero-shot zone assignment in a warehouse setting.
Trained policy outperformed CBS in all scenarios.
Achieved 79–95% success rate under dynamic obstacles compared to 0–14% for CBS with dynamic conditions.
Successfully accepted natural language commands for zone assignments.

Abstract

Multi-Agent Path Finding (MAPF) is critical for warehouse automation. Classical solvers like CBS provide optimal guarantees but cannot handle dynamic environments. Fully learned approaches suffer from cold-start. We present CBS-Bootstrapped MAPPO: a three-phase curriculum using CBS as a decaying reward signal to bootstrap MARL. The trained policy surpasses CBS at every difficulty level, maintains 79–95% success under dynamic obstacles (vs 0–14% for CBS+dynamic), and accepts natural language commands via a local LLM (qwen:4b via Ollama) for zero-shot zone assignment across a 20×20 warehouse grid with 12 agents.

AI से पूछें

Bookmark

View Full Paper

AI से पूछें

Bookmark

View Full Paper

CBS-Bootstrapped MAPPO: Scalable Multi-Agent Path Finding Under Dynamic Obstacles and Natural Language Control

Key Points

Abstract

Cite This Study

Also Consider

Also Consider