KD-MARL: Resource-Aware Knowledge Distillation in Multi-Agent Reinforcement Learning

2026-04-08

AI Summary

This work addresses the challenge of deploying high-performance multi-agent reinforcement learning systems on edge devices, where computational resources, memory, and inference latency constraints preclude the direct use of computationally expensive centralized expert policies. To overcome this limitation, the authors propose a two-stage knowledge distillation framework that transfers both action behaviors and coordination structures from the expert to lightweight decentralized student agents—without relying on critic networks. By leveraging advantage-based signal distillation and structured policy supervision, the method effectively captures the expert’s collaborative capabilities while accommodating heterogeneous agent architectures tailored to varying observation complexities. Evaluated on SMAC and MPE benchmarks, the approach retains over 90% of the expert’s performance while reducing FLOPs by up to 28.6×, substantially enhancing deployment efficiency in resource-constrained environments.

Abstract

Real world deployment of multi agent reinforcement learning MARL systems is fundamentally constrained by limited compute memory and inference time. While expert policies achieve high performance they rely on costly decision cycles and large scale models that are impractical for edge devices or embedded platforms. Knowledge distillation KD offers a promising path toward resource aware execution but existing KD methods in MARL focus narrowly on action imitation often neglecting coordination structure and assuming uniform agent capabilities. We propose resource aware Knowledge Distillation for Multi Agent Reinforcement Learning KD MARL a two stage framework that transfers coordinated behavior from a centralized expert to lightweight decentralized student agents. The student policies are trained without a critic relying instead on distilled advantage signals and structured policy supervision to preserve coordination under heterogeneous and limited observations. Our approach transfers both action level behavior and structural coordination patterns from expert policies while supporting heterogeneous student architectures allowing each agent model capacity to match its observation complexity which is crucial for efficient execution under partial or limited observability and limited onboard resources. Extensive experiments on SMAC and MPE benchmarks demonstrate that KD MARL achieves high performance retention while substantially reducing computational cost. Across standard multi agent benchmarks KD MARL retains over 90 percent of expert performance while reducing computational cost by up to 28.6 times FLOPs. The proposed approach achieves expert level coordination and preserves it through structured distillation enabling practical MARL deployment across resource constrained onboard platforms.

Problem

• multi-agent reinforcement learning
• knowledge distillation
• resource-constrained deployment
• heterogeneous agents
• coordination structure

Innovation

• Knowledge Distillation
• Multi-Agent Reinforcement Learning
• Resource-Aware
• Coordination Preservation
• Heterogeneous Agents