Ant Group's Lingbo Technology Releases Open-Source LingBot-Touch Tactile Foundation Model

Recently, general large models have continued to break through in capabilities such as task understanding and complex reasoning, and which technical route the robot “brain” should take has once again become the focus of discussion in the embodied intelligence industry: is it connecting to general large models and using digital intelligence to drive physical actions, or is it persisting with embodied native, taking physical perception and closed-loop operation as the foundation?

On November 10, Ant Lingbo Technology open-sourced the tactile foundation model LingBot-Touch, providing its own answer. The model is pre-trained based on nearly 5,000 hours of real tactile data, covering three mainstream sensing principles: magnetic tactile, visual tactile, and piezoresistive. It can convert different signals into the same set of tactile representations for use by different dexterous hands and motion models, allowing robots to further complement the tactile perception required for fine manipulation while they “see clearly and think clearly.”

Previously, Ant Group's Lingbo General Brain had already covered high-precision spatial perception, high-dynamic environment interaction, and generalizable dexterous manipulation, allowing robots to see more clearly, think more clearly, and work more efficiently. This time, LingBot-Touch has been connected to LingBot-VLA, extending perception capabilities to the “fingertips,” allowing robots to “touch more sensitively,” and further enhancing the embodied native general brain capability.

At the point of the accelerated evolution of general large models, choosing to launch a tactile foundation model oriented towards physical operation closed-loops conveys Ant Lingbo's continuous investment and determination in the embodied native route.

26 real-machine tests, task success rate increased by 14 to 26 percentage points

As general large models continue to progress, robots' abilities to understand complex tasks and reason through operation steps are being enhanced. However, in the real physical world, for every step of action a robot performs, it must perceive the changes occurring in the object and environment, and then correct the next step of action. In many delicate operations, a robot can see where the object is and can extend its hand to the correct position, but it may not necessarily be able to successfully complete the task, because after the fingers touch the object, whether the force is appropriate and whether the object is slipping are difficult to judge by vision alone, and require tactile information to provide supplements.

Taking a dexterous hand unscrewing a bottle cap as an example, depth information can help the fingers align with the bottle cap. After starting to rotate, the bottle cap may already be loose, or the fingertips may be sliding along the cap wall. These two situations are similar in the image, yet require different action adjustments. At this moment, the contact position of the fingertips, the magnitude of force and its changes, all become important judgment bases for the action model, while this information is often obstructed by the fingers and the object, making it difficult to obtain via vision.

To verify the help of tactile representation for real operations, Ant Lingbo conducted 26 real-machine tests on two dexterous hands, Sharpa Wave and XHand1, covering delicate operations such as turning book pages, unscrewing bottle caps, and identifying mahjong tiles. After connecting LingBot-Touch, the average task success rates of four action models—LingBot-VLA, π0.5, Diffusion Policy, and T-Rex—increased by approximately 14 to 26 percentage points compared to when no tactile information was used.

Taking T-Rex controlling Sharpa Wave to unscrew a bottle cap as an example, the task success rate without tactile information was 56.25%, the success rate with direct input of raw sensor signals was 62.5%, and the success rate reached 90% after connecting the tactile representation of LingBot-Touch.

(Figure caption: After connecting LingBot-Touch, the real-machine task success rates of the four action models on the two dexterous hands all reached their respective highest levels)

Unify tactile representation, allowing robot “hand feel” to be reused across hardware

Equipping a robot with tactile sensors does not mean it can use touch well. Currently, commercial tactile sensors as a whole are still in the early stages of development; mainstream routes such as visual touch, magnetic touch, and piezoresistive touch each have their own advantages and limitations, and multiple technical routes may coexist for a long time depending on different costs and usage scenarios.

What is more troublesome is that the data collected by these sensors is difficult to generalize directly. For the same contact, different raw signals will be formed under different sensing principles and contact point layouts. If a model only learns to interpret one type, after replacing the sensor or the dexterous hand, the existing data and tactile experience may be difficult to reuse, requiring re-collection, re-training, and re-adaptation. In the current era of rapid hardware iteration, this binding causes repetitive development, increasing the adaptation costs between hardware manufacturers, complete machine teams, and model developers.

Therefore, the key to the large-scale implementation of touch is to let tactile understanding break free from the dependence on specific sensor physical signals, enabling data from different sources to be used jointly for training, so that accumulated tactile capabilities can be reused across sensors, across dexterous hands, across models, and across tasks.

To this end, LingBot-Touch sets up encoding entries according to sensing principles, first processing different raw signals, and then mapping contact points of different layouts to a unified fingertip coordinate space, where a shared model learns the contact position, the force exerted, and their changes over time, outputting unified tactile Tokens. Consequently, data collected by humans wearing tactile gloves and teleoperated dexterous hands can be jointly used for training, and different action models can also use the trained model to extract contact features. This architecture also leaves expansion space for connecting new types of tactile sensors in the future.

(Figure caption: LingBot-Touch encodes magnetic touch, visual touch, and piezoresistive touch 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 pre-training data, and all magnetic tactile experience learned by the model came from data collected by humans wearing gloves. After switching to this dexterous hand, the team kept the LingBot-Touch parameters unchanged and only used the operation demonstrations of XHand to train the action model.

At the same time, LingBot-Touch integrated four different motion models in this real-machine test, and it is not bound to the self-developed model by Ant Lingbo. It provides general perception capabilities located between the tactile sensors and the motion models, rather than a dedicated module for a specific piece of hardware or a specific strategy. For the industry, sensor and dexterous hand manufacturers can choose hardware based on cost and task requirements, and the whole-machine team can also retain its own motion strategies, thereby reducing the repeated investment of re-developing tactile capabilities for every replacement of a hardware and algorithm combination.

To balance perception capabilities and edge-side computing power, Ant Lingbo provided two capacity models. This time, all 26 real-machine tests were completed using the lightweight version LingBot-Touch-Tiny with approximately 5.3 million parameters. After integrating LingBot-VLA, π0.5, and T-Rex, the model parameter count only increased by about 0.13% to 0.16%.

The larger capacity version, LingBot-Touch-Small, showed potential for further improvement in perception tests—when the pre-training data increased to nearly 5000 hours, the model could still benefit from the added data, indicating that there is still room for continuous expansion and improvement in tactile pre-training.

Adhering to embodied native, advancing in synergy with general large models

Regarding the recent evolution of embodied large model technology routes, Zhu Xing, CEO of Ant Lingbo, stated: “The progress of digital world large language models, represented by GPT-6 Astra and Claude Fable 5.2, has indeed made robots' complex task reasoning and physical operation capabilities stronger; we are paying close attention to this point and are actively absorbing it. However, practice on the industrial front tells us that to truly complete the operation loop, the hard bone of physical operation generalization still cannot be bypassed, and to a large extent, it still relies on the breakthroughs of the embodied foundation models themselves.”

In his view, the relationship between general large models and embodied foundation models is closer to a 1+1>2 synergy: the former enhances the understanding and reasoning of tasks and environments, while the latter improves physical perception and closed-loop operation capabilities; both break through individually and promote each other, jointly accelerating the maturity of the embodied intelligence industry.

“The quality of modal input largely determines the quality of action output; in other words, without good input, there is no good output,” said Zhu Xing. Ant Lingbo will persist in the embodied native route, continuously exploring new limits of physical operation generalization, creating a robot industry general brain across configurations, promoting the alignment of more physical modalities such as vision, depth, touch, and even temperature in a unified model, and continuously opening to the community as open source while advancing commercialization.

Currently, the technical report, model, and code of LingBot-Touch have been open-sourced, and developers can access and download them on Hugging Face, GitHub, and the ModelScope community.

Disclaimer: This article is reprinted from another website, aiming to provide readers with more news information. The content involved does not constitute investment or consumption advice, and is only for reference by readers.

【Editor: Shao Bingqi】