One Policy to Control Them All: Shared Modular Policies for Agent-Agnostic Control
This addresses the challenge of agent-agnostic control in reinforcement learning, enabling more flexible and efficient policy deployment across different robotic or simulated agents, though it is incremental as it builds on modular and decentralized approaches.
The paper tackled the problem of learning a single global reinforcement learning policy that can control diverse agent morphologies with varying state and action spaces, and showed that this policy successfully generates locomotion for planar agents like monopod hoppers, quadrupeds, and bipeds, generalizing to unseen variants without per-morphology training.
Reinforcement learning is typically concerned with learning control policies tailored to a particular agent. We investigate whether there exists a single global policy that can generalize to control a wide variety of agent morphologies -- ones in which even dimensionality of state and action spaces changes. We propose to express this global policy as a collection of identical modular neural networks, dubbed as Shared Modular Policies (SMP), that correspond to each of the agent's actuators. Every module is only responsible for controlling its corresponding actuator and receives information from only its local sensors. In addition, messages are passed between modules, propagating information between distant modules. We show that a single modular policy can successfully generate locomotion behaviors for several planar agents with different skeletal structures such as monopod hoppers, quadrupeds, bipeds, and generalize to variants not seen during training -- a process that would normally require training and manual hyperparameter tuning for each morphology. We observe that a wide variety of drastically diverse locomotion styles across morphologies as well as centralized coordination emerges via message passing between decentralized modules purely from the reinforcement learning objective. Videos and code at https://huangwl18.github.io/modular-rl/