MO-MIX: Multi-Objective Multi-Agent Cooperative Decision-Making With Deep Reinforcement Learning

Researchers propose MO-MIX, a new deep reinforcement learning approach that addresses multi-objective multi-agent cooperative decision-making problems. The method combines centralized training with decentralized execution and demonstrates superior performance over baseline methods while requiring less computational cost.

• →MO-MIX addresses the intersection of multi-objective and multi-agent reinforcement learning, an area with limited prior research.
• →The approach uses centralized training with decentralized execution framework with weight vectors for objective preferences.
• →A parallel mixing network architecture estimates joint action-value functions for cooperative decision-making.
• →The method includes an exploration guide to improve uniformity of non-dominated solutions in the Pareto set.
• →Experimental results show superior performance across all evaluation metrics with reduced computational requirements.