Ant Lingbo Open-Sources Tactile Foundation Model LingBot-Touch

On November 10, Ant Lingbo Technology open-sourced LingBot-Touch, a tactile foundation model that fills in the critical tactile perception capability for robot brains. The model was pretrained on nearly 5,000 hours of real tactile data and covers three mainstream sensing principles—magnetic tactile, visuotactile, and piezoresistive—and can convert different signals into a single set of tactile representations for use by different dexterous hands and action models, enabling robots to "see clearly and think clearly" while further filling in the tactile perception required for fine manipulation. At a time when general-purpose large models are accelerating, the choice to launch a tactile foundation model oriented toward the closed loop of physical manipulation conveys Ant Lingbo's continued investment in and commitment to the embodied-native route. Previously, Ant Lingbo's general-purpose brain already covered high-precision spatial perception, high-dynamic environment interaction, and generalizable dexterous manipulation, enabling robots to see more clearly, think more clearly, and act more deftly. This time, LingBot-Touch has been integrated into LingBot-VLA, extending perception capability to the "fingertips," enabling robots to "feel more sensitively" and further enhancing the capabilities of the embodied-native general-purpose brain. 26 real-robot tests: task success rates increase by 14 to 26 percentage points. As general-purpose large models continue to advance, robots' ability to understand complex tasks and reason about operation steps is strengthening. But in the real physical world, every step a robot executes requires perceiving changes in the object and environment and then correcting the next step. In many fine manipulation tasks, robots can see where the object is and can move their hands to the correct position, yet may not successfully complete the task, because after the fingers touch the object, whether the force is appropriate and whether the object slips are difficult to judge by vision alone and require tactile supplementation. Take tightening a bottle cap with a dexterous hand as an example. Depth information can help the fingers align with the cap. After rotation begins, the cap may already be loose, or the fingertips may be slipping along the cap wall. These two situations look similar in the image but require different action adjustments. At this point, the contact position at the fingertips, the magnitude of force, and their changes all become important bases for the action model's judgments, and this information is often occluded by the fingers and the object, making it difficult for vision to obtain. To verify the help of tactile representations for real manipulation, Ant Lingbo conducted 26 real-robot tests on two dexterous hands, Sharpa Wave and XHand1, covering fine manipulation tasks such as turning book pages, tightening bottle caps, and recognizing mahjong tiles. After integrating LingBot-Touch, the average task success rates of four action models—LingBot-VLA, π0.5, Diffusion Policy, and T-Rex—increased by about 14 to 26 percentage points compared with the no-tactile condition. Taking T-Rex controlling Sharpa Wave to tighten a bottle cap as an example, the task success rate was 56.25% without touch, 62.5% with raw sensor signals input directly, and reached 90% after integrating LingBot-Touch's tactile representations. After integrating LingBot-Touch, the real-robot task success rates of the four action models on both dexterous hands all reached their respective highest levels. Unified tactile representations let robots' "sense of touch" be reused across hardware. Installing tactile sensors on a robot does not mean it can make good use of touch. At present, commercial tactile sensors as a whole are still at a relatively early stage of development, and mainstream routes such as visuotactile, magnetic tactile, and piezoresistive each have advantages and limitations. For different costs and use scenarios, multiple technical routes may coexist for a long time. More troublesome is that the data collected by these sensors is difficult to use directly across the board. The same contact, under different sensing principles and contact layouts, will form different raw signals. If a model learns to interpret only one of them, then after changing sensors or dexterous hands, the existing data and tactile experience may be difficult to reuse, requiring re-collection, retraining, and re-adaptation. At a time when hardware is iterating rapidly, this binding causes repeated development and increases adaptation costs among hardware vendors, complete-machine teams, and model developers. Therefore, the key to scaling tactile technology is to free tactile understanding from dependence on the physical signals of specific sensors, so that data from different sources can be used together for training and accumulated tactile capabilities can be reused across sensors, dexterous hands, models, and tasks. To this end, LingBot-Touch sets up separate encoding entrances according to sensing principles, first processing different raw signals, then mapping contact points of different layouts into a unified fingertip coordinate space, where a shared model learns contact position, force, and their changes over time and outputs unified tactile Tokens. As a result, data collected by people wearing tactile gloves and by teleoperated dexterous hands can be used together for training, and different action models can also use the trained model to extract contact features. This architecture also leaves room for expansion to integrate new types of tactile sensors in the future. LingBot-Touch encodes magnetic tactile, visuotactile, and piezoresistive tactile signals into unified tactile Tokens for use by different action models. On XHand1, the team further verified the cross-hardware reuse of tactile capabilities. The magnetic tactile sensor model used by XHand1 was not in the pretraining data; all the magnetic tactile experience learned by the model came from data collected by people wearing gloves. After switching to this dexterous hand, the team kept LingBot-Touch's parameters unchanged and trained the action model only with XHand operation demonstrations. At the same time, LingBot-Touch was integrated with four different action models in this real-robot test and is not bound to Ant Lingbo's self-developed models. What it provides is general perception capability located between tactile sensors and action models, rather than a dedicated module for a particular hardware or policy. For the industry, sensor and dexterous hand manufacturers can choose hardware based on cost and task requirements, and full-robot teams can also retain their own motion policies, thereby reducing the repeated investment of redeveloping tactile capabilities every time a hardware and algorithm combination is changed. To balance perception capability and on-device compute, Ant Lingbo offers models in two capacities. All 26 real-robot tests this time were completed using the lightweight version LingBot-Touch-Tiny with about 5.3 million parameters. After integration with LingBot-VLA, π0.5, and T-Rex, the model parameter count increased by only about 0.13% to 0.16%. The larger-capacity version, LingBot-Touch-Small, showed potential for further improvement in perception tests—when pretraining data increased to nearly 5,000 hours, the model could still benefit from the newly added data, indicating that tactile pretraining still has room for continuous scaling and improvement. Staying true to embodied nativity and advancing in synergy with general large models. Regarding the recent evolution of embodied large model technology routes, Ant Lingbo CEO Zhu Xing said: "Progress in digital-world large language models represented by GPT-6 Astra and Claude Fable 5.2 has indeed made robots' complex task reasoning and physical manipulation capabilities stronger, and we are paying close attention to this and actively absorbing it. But practice on the industry front line tells us that to truly close the manipulation loop, the hard nut of physical manipulation generalization remains unavoidable, and to a large extent it still depends on breakthroughs in embodied foundation models themselves." In his view, general large models and embodied foundation models are closer to a 1+1>2 synergistic relationship: the former enhances understanding of and reasoning about tasks and environments, while the latter improves physical perception and closed-loop manipulation capabilities. The two make breakthroughs in their respective areas and promote each other, jointly accelerating the maturation of the embodied intelligence industry. "The quality of modality input largely determines the quality of action output; in other words, without good input there is no good output," Zhu Xing said. Ant Lingbo will adhere to the embodied-native route, continue exploring new upper limits of physical manipulation generalization, build a general-purpose brain for the robotics industry across configurations, promote the alignment of more physical modalities such as vision, depth, touch, and even temperature in a unified model, and continue to open source to the community while advancing commercialization. Currently, the technical report, model, and code for LingBot-Touch have been open-sourced, and developers can access and download them on Hugging Face, GitHub, and the ModelScope community.