Learning Whole-Body Control for a Salamander Robot
This work addresses the challenge of deploying learned controllers on highly articulated amphibious robots, which is incremental as it applies existing reinforcement learning methods to a new domain.
The researchers tackled the problem of learning joint-level whole-body control for a salamander robot that transfers from simulation to hardware, achieving stable and coordinated walking on various terrains in the real world. They also demonstrated transitions between walking and swimming in simulation.
Amphibious legged robots inspired by salamanders are promising in applications in complex amphibious environments. However, despite the significant success of training controllers that achieve diverse locomotion behaviors in conventional quadrupedal robots, most salamander robots relied on central-pattern-generator (CPG)-based and model-based coordination strategies for locomotion control. Learning unified joint-level whole-body control that reliably transfers from simulation to highly articulated physical salamander robots remains relatively underexplored. In addition, few legged robots have tried learning-based controllers in amphibious environments. In this work, we employ Reinforcement Learning to map proprioceptive observations and commanded velocities to joint-level actions, allowing coordinated locomotor behaviors to emerge. To deploy these policies on hardware, we adopt a system-level real-to-sim matching and sim-to-real transfer strategy. The learned controller achieves stable and coordinated walking on both flat and uneven terrains in the real world. Beyond terrestrial locomotion, the framework enables transitions between walking and swimming in simulation, highlighting a phenomenon of interest for understanding locomotion across distinct physical modes.