Agile perceptive multi-skill locomotion for quadrupedal robots in the wild

Researchers developed APT-RL, a framework enabling quadrupedal robots to learn agile locomotion skills from trajectory-optimization data and apply them via reinforcement learning on complex terrain. The system generated 180,000 trajectories in 8 minutes, demonstrating rapid acquisition of reusable locomotion representations for real-world deployment.

APT-RL first learns reusable locomotion representations from trajectory-optimization data and then uses these representations as priors for reinforcement learning on complex terrain. Trajectory optimization based on single rigid body dynamics generated 180,000 trajectories 15.5 hours of motion in 8 minutes . The dataset contains both state trajectories and their corresponding control inputs.