robot-atlas

Domain

RL, Sim-to-Real & Locomotion

Reinforcement learning at scale, massively parallel simulation, and the transfer problem.

6 of 6 modules published

  1. The MDP simulability gap: contact-rich manipulation resists the simulation that made walking routine.

  2. Isaac Lab, Newton, MJX, and Brax: GPU-parallel environments and the wall-clock economics of training.

  3. Domain randomization, teacher-student distillation, system identification, and real-to-sim correction.

  4. From ANYmal to Unitree and the MIT humanoid line: how learned gaits became the default.

  5. Motion tracking from PHC to ASAP and GMT, and the three decompositions of 2026.

  6. LLM-written rewards, curricula, and where classical trajectory optimization still wins.