机器人学习与Manipulation研发科学家_CR

Develop end-to-end visuomotor policies for humanoid and general-purpose robots, enabling perception-to-action learning. Engineer manipulation capabilities including grasping, transport, object reorientation, door opening, and dual-arm assembly using Behavior Cloning, Reinforcement Learning, and VLA methods. Build robust strategies resilient to sensor noise, partial observability, and contact dynamics by managing the full technical chain from real-world data collection and dataset construction to model training, evaluation, and deployment. Leverage simulation and digital twins for scalable training and Sim-to-Real transfer, prioritizing actual robot performance. Collaborate with perception, base models, planning, control, and hardware teams to integrate learned strategies into complete robotic systems. Track Embodied AI and Robot Learning frontiers to drive research into production.