Learning to Control Self-Assembling Morphologies: A Study of Generalization via Modularity
This addresses the challenge of adaptive control in robotics and AI by enabling agents to form and control complex bodies dynamically, though it is incremental as it builds on modular and evolutionary approaches.
The paper tackles the problem of controlling self-assembling morphologies by developing a modular co-evolution strategy where primitive agents learn to dynamically assemble into composite bodies and coordinate behavior, demonstrating better generalization to test-time changes in environment and agent structure compared to static baselines.
Contemporary sensorimotor learning approaches typically start with an existing complex agent (e.g., a robotic arm), which they learn to control. In contrast, this paper investigates a modular co-evolution strategy: a collection of primitive agents learns to dynamically self-assemble into composite bodies while also learning to coordinate their behavior to control these bodies. Each primitive agent consists of a limb with a motor attached at one end. Limbs may choose to link up to form collectives. When a limb initiates a link-up action, and there is another limb nearby, the latter is magnetically connected to the 'parent' limb's motor. This forms a new single agent, which may further link with other agents. In this way, complex morphologies can emerge, controlled by a policy whose architecture is in explicit correspondence with the morphology. We evaluate the performance of these dynamic and modular agents in simulated environments. We demonstrate better generalization to test-time changes both in the environment, as well as in the structure of the agent, compared to static and monolithic baselines. Project video and code are available at https://pathak22.github.io/modular-assemblies/