Research Lab - Spatial Intelligence

💡 We believe the above three thrusts are deeply synergistic: cognitive models trained in simulation are grounded in physical robots, whose real-world experiences in turn benefit both the cognitive models and digital worlds. Our long-term vision is to build a self-sustaining, self-evolving spatial AI ecosystem where machines autonomously perceive, reason about, and transform the physical world—enabling AI Designs AI, Robots Build Robots through the seamless convergence of embodied cognition, digital universe, and physical embodiments.

💪 Towards this long-term vision, we are currently pushing the following research directions:

• Prototyping spatially-grounded foundation models
• Exploring human-like efficient spatial representations
• Building open-world embodied AI simulators
• Grounding visual-spatial intelligence in humanoid robots
• Constructing large-scale datasets and benchmarks to broaden the practical applications of spatial AI
• Designing specialized physical embodiments to diversify problem domains of spatial AI

📣 To advance our mission, we are building an interdisciplinary team and welcome researchers with diverse expertise, including but not limited to: (1) AI Computing (e.g., MLLMs, UMMs, VFMs, world model, generative model), (2) 3D Vision (e.g., 3DGS, NeRF, SLAM), and (3) Robotics (e.g., sim2real, humanoid robotics, field robotics, robotic design). Please send me an email, and I will get back to you if there is a good fit!