RSS jumping-control paper transfers versatile dynamic jumps to Cassie hardware

Key points

• The RSS paper trains a reinforcement-learning controller for robust bipedal jumping across different distances and directions.
• After multi-stage training, the policy transfers directly to real Cassie hardware for standing long jumps, elevated-platform jumps, and multi-axis jumps.
• It is a strong Cassie research result because it pushes the platform beyond walking into aggressive dynamic maneuvers.