Embodied AI is the integration of artificial intelligence with the physical world, enabling robots to interact with and learn from the real world. We focus on the most critical areas of embodied AI, including humanoid, robot manipulation, and dexterous hand. Our goal is to explore the scaling law for robots, develop general world models, and unveil the power of reinforcement learning to achieve general-purpose embodied agents.
Autonomous Driving stands at the intersection of intelligence, world modeling, and safety alignment, enabling vehicles to respond to the surroundings effectively for both comfort and safety. We target the crucial areas of autonomous driving, including whole-scene perception systems, critical data generation, and end-to-end decision-making. Our mission is to establish a comprehensive pipeline by leveraging massive real-world driving data and building efficient world representation for safe and generalizable autonomy.
