机器人基座模型研发科学家_CR

Conduct research on next-generation robot base models, world models, and world action models to advance general embodied intelligence. Develop the Vision-Geometry-World-Action paradigm by jointly learning visual representations, 3D geometry, temporal dynamics, physical world representations, and robot actions. Utilize large-scale multimodal data to learn generalized physical world representations and map environment understanding directly to robot actions. Explore relationships between world modeling, future prediction, and action generation, including training-stage physical knowledge learning and inference-stage efficient action generation. Investigate unsupervised video learning, multimodal pretraining, imitation learning, and reinforcement learning for scalable training. Construct large-scale robot video-action datasets and training pipelines, managing data governance and scaling across robot bodies. Validate models on simulation and real robot platforms covering robotic arms, mobile robots, humanoid robots, and industrial robots. Track Physical AI and robotics foundation model frontiers to drive research into Bosch technology platforms and future products. Collaborate with Bosch China and global teams on technical strategy, prototyping, technology transfer, and partner acquisition.