Robot body and movement co-designed for better walking in rough places
GLAMDRING: Gait Learning And Morphology co-Design via Reinforcement LearnING of CPGs
RoboticsMachine Learning
Summary
Robots need to move well in tough environments like disaster zones or farms, but it's hard to design both their bodies and how they walk at the same time. This work presents GLAMDRING, a system that designs a robot’s shape and figures out how it should walk together, making them work best for each other. The authors show that designing a robot’s body and movement together is important to meet goals like speed and carrying stuff, and that common animal walking patterns naturally emerge from this process. They also tested a real robot to prove the idea works outside of computer simulations.
What this means in practice
- •For robot developers: Design quadruped robots with jointly optimized body and walking controllers tailored for specific tasks and environment constraints.
- •For disaster response teams: Deploy robots whose body shape and walking gaits are customized to carry required loads and move efficiently in complex, unstructured environments.
Authors
Amogh Joshi, Kaushik Roy
Abstract
Robots are moving out of the structured factory floor and into unstructured environments such as disaster sites, planetary surfaces, and agricultural fields, for which the right robot often does not yet exist. We present GLAMDRING, a framework that synthesizes the optimal robot for a locomotion task and, jointly, learns the controller that drives it. For the given specifications of forward-velocity bounds, a per-actuator power budget, an actuator library, and a payload requirement, GLAMDRING returns a matched quadruped morphology (link geometry and per-joint actuators) and a Hopf-oscillator Central Pattern Generator (CPG) gait policy. We rank feasible designs against a target design objective, viz., maximum speed, minimum Cost of Transport (CoT), or max Payload Margin. Because body and locomotion are coupled, the optimal morphology dictates how a robot is driven, while optimal gait depends on the physical body. We train a small number of CPG policies by reinforcement learning across the space of candidate morphologies, co-learning the gait with the underlying robot hardware. Link lengths and actuators are then resolved post-hoc from the policy's logged operating envelope, reducing synthesis cost to a small, fixed number of reinforcement-learning runs instead of one per candidate. Our experiments show three key findings: co-designing body and gait is necessary to satisfy locomotion constraints; actuator-envelope feasibility, rather than locomotion success alone, determines realizable payload capacity; and canonical animal gaits emerge naturally in most designs from morphology and constraints alone. A real-world demonstration further highlights the efficacy of our work.