A robot can recognize an egg on the table, but it cannot necessarily pick it up. This is the real gap facing embodied intelligence today: vision and decision-making models are advancing rapidly, but capabilities in the "hand" — especially the perception of contact, pressure, and slip — remain a weak point. According to an announcement, LDROBOT (01236) released a voluntary announcement on the evening of October 11, stating that the company has recently entered into a strategic cooperation with Huazhong University of Science and Technology (HUST), and will jointly conduct research and development with Professor Li Kan's team on flexible three-dimensional tactile sensors and array sensing systems for embodied robots. A robotics company that started with "eyes" (LiDAR, visual perception) is now betting its next step on "skin."

The bottleneck of embodied intelligence is shifting from "seeing" to "doing"

Over the past two years, large models have rapidly evolved the robot's "brain," and visual perception solutions have also matured. But for robots to move from showrooms into factories and homes, they must complete real physical interactions such as grasping, assembly, and collaboration. At this point, vision is no longer sufficient: occluded objects, fragile items, smooth-surfaced parts, and force control requiring precision all exceed what vision can determine. When a human completes a grasp, it relies on real-time feedback from the fingertips regarding pressure, shear, and slip. Without this layer of feedback, robots can only "blindly grasp by sight," either failing to hold securely or applying excessive force. The industry has therefore gradually reached a consensus that vision determines what a robot "knows," while touch determines what a robot "can do." For dexterous hands and humanoid robots to move from demonstration to practical use, touch is an unavoidable link.

Why "three-dimensional"? This is the watershed of tactile perception

Tactile sensing is not a new concept, but most existing solutions can only perceive "how much pressure is applied," namely the normal force perpendicular to the contact surface. In real manipulation, however, more critical information lies in the other two directions: the tangential forces along the contact surface in the forward-backward and left-right directions, which determine whether an object is slipping and whether a finger is being dragged. Only by simultaneously acquiring forces in all three directions can a robot judge whether its grip is stable and then adjust force and posture. This is why 3D tactile sensing has become a focus of industry research — it is the data foundation for dexterous grasping, fine assembly, and human-robot collaboration. But the difficulty is equally high. Sensing units must be soft enough to conform to curved surfaces while withstanding repeated contact, and the force signals in three directions must be precisely decoupled through algorithms. As array density increases, the difficulty of synchronous acquisition and packaging also multiplies. If any link in materials, microstructures, packaging, or algorithms fails, mass production becomes impossible.

Demand has been validated, standards have not yet formed, and the tactile track is in a window period

From an industry progress perspective, 3D tactile sensing is currently at a typical early stage: demand has been validated, but technical routes and industry standards have not yet converged. Demand has already emerged on the data acquisition side, but large-scale volume on the execution side still takes time. More noteworthy is that mainstream embodied intelligence large models have not yet incorporated tactile modalities at scale. Once leading models begin systematically introducing tactile data, hardware supply capacity will quickly become a bottleneck. Therefore, the truly valuable time gap appears before large models adopt tactile sensing. Whoever can first build sensors, make them stable, and achieve volume will hold the early advantage in data accumulation and engineering experience. According to a report by China Insights Consultancy, other perception technologies including hearing, touch, and force sensing are expected to maintain a compound annual growth rate of 20.8% from 2024 to 2029, with the global market size reaching RMB 34 billion by 2029. The track is still small in scale, but it is on the eve of an outbreak.

A reasonable extension for a "perception company"

Against this backdrop, LDROBOT's choice of 3D tactile sensing is not surprising. The company has long accumulated perception software and hardware around LiDAR, SLAM/VSLAM, and AI scene recognition, forming a dual-wheel-driven model of "visual perception products + complete robots," and has completed commercial validation in scenarios such as robotic lawn mowers. Extending from "seeing" to "touching" is a natural step for a perception company expanding into multimodality. According to LDROBOT's announcement, the two parties will rely on the State Key Laboratory of Intelligent Manufacturing Equipment and Technology at Huazhong University of Science and Technology to carry out research, development, and scenario validation of flexible 3D tactile sensor devices and array systems around the tactile perception needs of embodied robots. Specifically, this includes three directions: developing flexible 3D tactile sensors and high-density sensing arrays to improve perception capability and long-term stability under complex working conditions; developing supporting signal acquisition hardware, 3D force decoupling, and tactile imaging algorithms to form an overall solution from sensor devices to array systems; and validating scenario implementation capability on platforms such as dexterous hands, robotic arms, humanoid robots, and collaborative robots. Professor Li Kan, who is leading this collaboration, is currently a doctoral supervisor at Huazhong University of Science and Technology and a recipient of the National Natural Science Foundation of China Excellent Young Scientists Fund (Overseas). He has long been devoted to research on 3D flexible structures and stretchable electronic devices, and has published multiple papers as first author or co-first author in journals such as Nature, Nature Communications, Science Advances, and PNAS. From the structure of the cooperation, this is a typical industry-academia-research division of labor. 3D flexible touch spans flexible electronics, specialty materials, precision packaging, and force decoupling algorithms. The threshold for basic research is high, requiring university teams to continue tackling frontier scientific problems. For sensors to truly become products, they also need repeated validation in complete-machine scenarios, which is precisely where LDROBOT's participation adds value. Such cooperation aims to bridge the "valley of death" between principle prototypes and mass-producible products.

From a business implementation path perspective, 3D tactile sensing can serve as an independent perception product line for customers such as dexterous hands, robotic arms, humanoid robots, and collaborative robots. It can also first be validated and implemented on LDROBOT's own complete-machine products, forming a differentiated hardware advantage. For the capital market, this is an extension of LDROBOT's "perception infrastructure" narrative toward multimodality. As the robotics industry moves from technical validation to large-scale application, upstream perception components often benefit first from category expansion and shipment growth. The company not only provides perception products to different types of robots but also promotes technology implementation through its own complete machines, giving its business structure a certain collaborative flexibility. However, 3D tactile sensing still takes time to move from the laboratory to mass production, and R&D progress and commercialization pace still need continuous tracking. In early September this year, the company was included in the Stock Connect program, and in October its share price once rose from HKD 23.5 to HKD 31.68, with a maximum gain of 34.8%. Investors can pay attention to the company's subsequent R&D progress and order implementation.