Touch as the gateway, data as the engine, foundation models at the core — powering Physical AI.
High-Resolution Multimodal Tactile Sensing
An advanced multidimensional visuo-tactile sensor delivers high-resolution tactile perception, enabling robots to perceive the physical world with sensitivity beyond human touch.
Real-World Physical Interaction Datasets
A large-scale multimodal physical interaction dataset with tactile information is powered by a full-stack data infrastructure. Continuously expanding real-world interaction data drives model iteration and advances robotic intelligence.
Tactile Physical World Foundation Model
The tactile physical world foundation model combines the proprietary VTLA architecture with multimodal alignment and cross-modal reasoning to enable causal understanding and dexterous manipulation for real-world robotic applications.
Applications
Embodied interactive data lays a solid foundation for model training.
Applications
Embodied interactive data lays a solid foundation for model training.
Applications
Embodied interactive data lays a solid foundation for model training.
Applications
Embodied interactive data lays a solid foundation for model training.
Applications
Embodied interactive data lays a solid foundation for model training.
Applications
Embodied interactive data lays a solid foundation for model training.
Applications
Embodied interactive data lays a solid foundation for model training.