Ant Lingbo Open-Sources Tactile Foundation Model LingBot-Touch

Ant Lingbo LingBot-Touch tactile foundation model open-sourced, giving the general-purpose brain a "sense of touch"

2026-11-10 10:46 IT Home - Haomiao

IT Home, November 10: Today, Ant Lingbo Technology open-sourced the tactile foundation model LingBot-Touch, filling in the key tactile perception capability for robot brains.

According to the official introduction learned by IT Home, pre-trained on nearly 5,000 hours of real tactile data, LingBot-Touch covers three mainstream sensing principles: magnetic tactile, visuotactile, and piezoresistive. It can convert different signals into a single set of tactile representations, allowing the accumulated "sense of touch" to be reused across sensors, dexterous hands, models, and tasks. In 26 real-machine tests, after integrating this representation, the average task success rate of four action models increased by about 14 to 26 percentage points compared with no tactile sensing.

Previously, the Ant Lingbo LingBot series models had already built a full-stack technical system covering high-precision spatial perception, high-dynamic environment interaction, and generalizable dexterous manipulation, enabling robots to see more clearly, think more intelligently, and work more efficiently. This time, LingBot-Touch extends the general-purpose brain's perception capability to the "fingertips," making robots "feel more sensitively" and further enhancing the embodied-native general-purpose brain capability.

After seeing clearly, why is "sensitive touch" still needed?

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 operations, seeing the object's position and reaching the hand to the right place does not mean the task can be completed smoothly. After contact, whether the force is appropriate and whether the object slips are often difficult to judge by vision alone.

Take unscrewing a bottle cap as an example. Depth information can help the fingers align with the cap. But after starting to turn, the cap may already be loose, or the fingertip may be slipping along the cap wall. These two situations look similar in the image but require different action adjustments. The contact position of the fingertip, the magnitude of force, and their changes become important judgment bases for the action model, 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 operations, the team conducted 26 real-machine tests on two dexterous hands, Sharpa Wave and XHand, covering fine operations such as turning book pages, unscrewing bottle caps, and identifying mahjong tiles. After integrating LingBot-Touch, the average task success rate of four action models—LingBot-VLA, π0.5, Diffusion Policy, and T-Rex—increased by about 14 to 26 percentage points compared with no tactile sensing.

Among them, in the task of T-Rex controlling Sharpa Wave to unscrew a bottle cap, the success rate was 56.25% without tactile sensing, 62.5% when directly inputting raw sensor signals, and reached 90% after integrating the LingBot-Touch tactile representation.

This series of real-machine experiments shows that high-quality tactile information input can significantly improve robot action success rates.

Unified tactile representation, reusable across dexterous hands and models

Equipping a robot with tactile sensors does not mean it can use touch well. Commercial tactile sensors as a whole are still at an early stage of development, and routes such as visuotactile, magnetic tactile, and piezoresistive each have advantages and limitations. For different costs and usage scenarios, multiple routes may coexist for a long time.

More troublesome is that the same contact, under different sensing principles and contact layouts, will form different raw signals. If a model only learns to interpret one of them, after changing sensors or dexterous hands, the existing data and tactile experience may be difficult to reuse, requiring re-collection, training, and adaptation.

The team's approach 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 the accumulated tactile capability can be reused across dexterous hands and action models.

To this end, LingBot-Touch sets up encoding entrances separately according to sensing principles, first processing different raw signals, and then mapping contacts of different layouts into a unified fingertip coordinate space. The shared model learns contact position, force, and their changes over time, and outputs unified tactile Tokens—tactile representations for use by different action models.

Under this architecture, 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.

On XHand, the team further verified the cross-hardware reuse of tactile capability. The magnetic tactile sensor model used by XHand is not in the pre-training data; all the magnetic tactile experience learned by the model comes from data collected by people wearing gloves. After switching to this dexterous hand, the team kept the LingBot-Touch parameters unchanged and only used XHand's operation demonstrations to train the action model.

LingBot-Touch provides a general perception capability located between tactile sensors and action models, rather than a dedicated module for a certain hardware or a certain policy. This test integrated four different action models and is not bound to the self-developed LingBot-VLA model. Sensor and dexterous hand manufacturers can choose hardware according to cost and task requirements, and complete-machine teams can retain their own action policies, reducing repeated investment in redeveloping tactile capability when changing hardware and algorithm combinations.

To balance perception capability and on-device computing power, the team provides two model capacities. All 26 real-machine tests this time were completed using the lightweight version LingBot-Touch-Tiny with about 5.3 million parameters. After integrating 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 pre-training data increased to nearly 5,000 hours, the model could still benefit from the newly added data, indicating that tactile pre-training still has room for continuous expansion and improvement.

Ant Lingbo stated that at a time when general large models are accelerating their evolution, the release of LingBot-Touch is precisely the team's continued investment in and commitment to the "embodied-native" route—unlike multimodal paradigms oriented toward digital information, robots are in the real physical world, with information coming from sensors and acting on the environment. They need not only to understand spatial perception information such as depth, but also to process contact perception modalities such as touch, force, and temperature. The quality of modality input largely determines the quality of action output.

Reference

Project homepage

Advertising statement: External redirect links contained in the article (including but not limited to hyperlinks, QR codes, passwords, and other forms) are used to convey more information and save selection time. The results are for reference only. All articles on IT Home containing external links include this statement.