Back to Open Positions

Robotics RL Motion Control Engineer

Shanghai

Responsibilities

  1. Develop reinforcement learning based robot motion control algorithms, including but not limited to quadruped/humanoid gait generation, whole-body control, and locomotion-manipulation coordination
  2. Design, implement, and optimize RL algorithms (such as PPO, SAC, TD3, Dreamer, Diffusion Policy) to address high-dimensional continuous control, sparse rewards, low sample efficiency, and the sim-to-real gap
  3. Complete sim-to-real transfer, combining system identification, adaptive control, residual learning, or online fine-tuning to ensure stability and robustness of algorithms on real hardware
  4. Work closely with hardware, perception, and planning teams to integrate RL controllers into complete robot systems, supporting real-robot deployment and testing
  5. Track the latest academic and industrial progress (ICRA, RSS, CoRL, NeurIPS, ICML, etc.), continuously improve algorithm performance, and produce technical reports or papers

Requirements

  1. Master's degree or above in Robotics, Control Science and Engineering, Computer Science, Artificial Intelligence, Automation, or related fields (outstanding Bachelor candidates may be considered)
  2. Solid theoretical foundation in reinforcement learning; familiar with mainstream on-policy/off-policy algorithms, Actor-Critic, model-based RL, and imitation learning
  3. Proficient in Python, familiar with PyTorch (or JAX), able to independently implement and debug deep RL algorithms
  4. Experience in robot motion control; understanding of kinematics/dynamics modeling and traditional control methods (PID/MPC/whole-body control) and able to combine them with RL
  5. Familiar with at least one mainstream physics simulator (MuJoCo, Isaac Gym, Gazebo, Webots, etc.); experience with a complete training-to-deployment loop is preferred