Learning Energy-Efficient Air--Ground Actuation for Hybrid Robots on Stair-Like Terrain
This addresses energy efficiency for hybrid robots on discontinuous terrain, though it appears incremental as it builds on existing RL and hybrid robot concepts.
The researchers tackled the problem of hybrid aerial-ground robots struggling with stair-like terrain due to energy inefficiency, proposing an energy-aware reinforcement learning framework that coordinates propellers, wheels, and tilt servos. The result was a policy achieving about 4 times lower energy than propeller-only control in simulation and 38% lower average power than a rule-based controller on hardware.
Hybrid aerial--ground robots offer both traversability and endurance, but stair-like discontinuities create a trade-off: wheels alone often stall at edges, while flight is energy-hungry for small height gains. We propose an energy-aware reinforcement learning framework that trains a single continuous policy to coordinate propellers, wheels, and tilt servos without predefined aerial and ground modes. We train policies from proprioception and a local height scan in Isaac Lab with parallel environments, using hardware-calibrated thrust/power models so the reward penalizes true electrical energy. The learned policy discovers thrust-assisted driving that blends aerial thrust and ground traction. In simulation it achieves about 4 times lower energy than propeller-only control. We transfer the policy to a DoubleBee prototype on an 8cm gap-climbing task; it achieves 38% lower average power than a rule-based decoupled controller. These results show that efficient hybrid actuation can emerge from learning and deploy on hardware.