Daolin Ma Positions Tactile Intelligence as Physical AI’s New Backbone

The transition from theoretical AI to physical deployment hinges on a machine’s ability to manipulate its surroundings rather than merely observing them. Ma contends that the next evolution of Physical AI requires a shift in focus from large-scale language modeling to the integration of comprehensive physical perception. By equipping robots with the capacity to interpret contact and deformation, developers can bridge the gap between intelligence and real-world utility.

Founded in May 2024, Shanghai-based Xense Robotics has centered its research on this tactile foundation. The company maintains a full-stack ecosystem that integrates multimodal tactile sensors, the proprietary XTac data collection system, and the VTLA foundation model. This architecture combines visual, linguistic, and tactile inputs to refine how robots handle physical tasks. With backing from investors like Hillhouse Ventures and Li Auto, Xense currently supplies technology to over 300 research institutions and industrial partners, including CATL and L'Oréal, aiming to standardize tactile intelligence as a core layer for the future of embodied AI.