Xense Robotics raises hundreds of millions of yuan for robot tactile sensors
What's the deal? Xense Robotics, a Shanghai-based startup building tactile sensors for robots, has raised a funding round of hundreds of millions of yuan. The company will use the capital to industrialise its optical-tactile sensors, deploy data collection equipment, and refine its embodied AI models.
What's the endgame? Xense is building a software-hardware system around "perception-data-model-scenario," aimed at letting robots recognise contact states, perceive forces, and adjust their actions in real-world settings. Its focus is shifting from tactile-perception research to sensor manufacturing, data infrastructure, and model validation.
The product: On hardware, Xense is accelerating production of its high-precision three-color optical-tactile sensor, which captures deformation, texture, slippage, and local geometry during object contact. That data feeds fine-grained grasping and contact judgment. The company has not disclosed production volumes, facility scale, or mass-production timelines.
What else? On data, Xense will deploy and validate its XTac UMI G1 collection system to build large-scale tactile datasets from robot-object interactions. On models, it is refining its VTLA multimodal system, which combines vision, touch, language, and action so robots can interpret tasks using both visual and contact feedback. No specific clients or deployment projects have been named.
Who's behind it? Founded in May 2024, Xense is led by Ma Daolin, an associate professor at Shanghai Jiao Tong University. Its portfolio spans multimodal tactile sensors, data collection systems, the VTLA embodied model, and a tactile world model.
The signal: Touch remains a weak point for robots that already handle vision and language well. By pairing sensors with proprietary datasets and models, Xense is betting that tactile feedback is the missing layer for embodied intelligence — and that owning the full stack, from hardware to data to model, is the path to commercialising it.
Read more: WeDoAny
