Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper addresses the challenge of bandwidth constraints in multi-agent reinforcement learning (MARL) by decoupling the communication pathway from the policy's latent representation. They introduce a normalized bandwidth budget, β, to unify different communication constraints and propose SLIM, a minimal architecture that isolates the impact of bandwidth limitations on performance. Experiments on partially-observable MARL benchmarks demonstrate that SLIM achieves state-of-the-art performance and exhibits robustness under limited communication, with only marginal performance degradation as bandwidth decreases.
Decoupling communication from policy in MARL lets you drastically reduce bandwidth needs without sacrificing performance, opening the door to real-world deployments with drone swarms and other bandwidth-limited systems.
Communication enables coordination in multi-agent reinforcement learning (MARL), but many real-world applications, e.g., search-and-rescue with drone swarms, operate under severe bandwidth constraints. Many communication architectures still expose a coupled bottleneck in which a shared latent representation is used for both policy execution and inter-agent communication. Consequently, reducing message size directly limits the policy's latent space, often leading to significant performance degradation. We address this with two contributions. First, we introduce β, a normalised per-agent bandwidth budget that unifies sparsity, rounds, and message dimension into a single comparable constraint. Second, we provide SLIM, a minimal architecture that decouples the communication pathway from the policy's latent representation, allowing us to isolate the effect of bandwidth from the effect of policy capacity while benefiting from in-step communication. We evaluate our method on several partially-observable MARL benchmarks, where communication is essential. Our approach achieves state-of-the-art performance and exhibits scalability and robustness under limited communication, with only marginal degradation as bandwidth is reduced.