• Design, implement, and evaluate RL algorithms for robotic control, motion planning, and adaptive behaviors in dynamic, unstructured environments.
• Develop and integrate RL policies with robot control systems, ensuring compatibility with hardware constraints and real-time requirements.
• Collaborate with perception teams to fuse RL with vision, depth, and sensor data for robust decision-making.
• Build and maintain sim-to-real pipelines, including domain randomization and transfer learning techniques.
• Conduct experiments on physical robots, including designing safety protocols and monitoring for unexpected behaviors.
• Leverage simulation environments (Isaac Gym, Gazebo, MuJoCo, PyBullet) for large-scale training before real-world validation.
• Continuously improve model efficiency to operate within compute and latency constraints on embedded robotic systems.