One Policy to Control Them All: Shared Modular Policies for Agent-Agnostic Control

TL;DR

This paper introduces Shared Modular Policies (SMP), a neural network approach enabling a single control policy to generalize across diverse agent morphologies by using local modules with message passing.

Contribution

The paper proposes a novel modular neural network architecture that generalizes control policies across various agent structures without retraining for each morphology.

Findings

Single modular policy controls multiple agent types

Emergence of diverse locomotion styles

Generalizes to unseen morphologies

Abstract

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/