Role Overview
We are seeking a Robotic Software Engineer specializing in Reinforcement Learning (RL) for Locomotion Control to develop advanced motion control strategies for legged robots. This role focuses on designing, training, and deploying RL-based controllers to enable agile, adaptive, and efficient robot locomotion in real-world environments. You will work at the intersection of robot control, reinforcement learning, and physics-based simulation, contributing to cutting-edge autonomous robotic systems.
Key Responsibilities
- Develop reinforcement learning-based locomotion controllers for legged robots
- Develop efficient sim-to-real transfer strategies to deploy trained policies on physical robots.
- Collaborate with hardware and perception teams to ensure smooth deployment of locomotion policies.
Requirements
- Education: Bachelors, Master’s or PhD in Robotics, Control Engineering, Computer Science, or a related field.
- Experience: Prior work in robotic locomotion, reinforcement learning, and optimal control.
- Experience training and deploying RL policies for legged or wheeled robots.
- Familiarity with whole-body control, torque control, and contact-aware planning.
Technical Skills
- Programming: Proficiency in Python and C++ for RL training and real-time control.
- Reinforcement Learning: Experience with PPO, SAC, TD3, DDPG, or custom RL approaches.
- Control Theory: Strong understanding of MPC, LQR, Whole-Body Control (WBC), and PID tuning.
- Simulation: Hands-on experience with Isaac Gym or MuJoCo or RaiSim, or Bullet Physics.
- Deep Learning & Optimization: Familiarity with PyTorch or TensorFlow.
- Robotics Frameworks: Experience with ROS 1/2, Gazebo