PPF paper improves Digit locomotion with model-assumption-based regularization

Key points

• PPF combines controller imitation, reinforcement-learning fine-tuning, and model-assumption-based regularization for humanoid locomotion.
• The paper reports hardware experiments on full-size Digit, including 1.5 m/s walking and robust locomotion on slippery, sloped, uneven, and sandy terrain.
• The RA-L publication gives Digit another source-backed result for learning-based locomotion under difficult real-world conditions.