Spot RL paper deploys high-speed learned locomotion on hardware

Key points

• The paper describes high-performance reinforcement learning deployment on Boston Dynamics Spot using the Spot RL Researcher Development Kit.
• The authors optimize simulation parameters with distributional measures and deploy policies on physical Spot hardware.
• The work gives Spot a public benchmark for high-speed learned locomotion, sim-to-real tuning, and low-level motor access research.