Yaole Technology Launches Integrated Knitted Data Glove, Bringing Tactile Capture Back to Real Human Hand Motions | Frontline

Author ｜ Huang Nan

Editor ｜ Yuan Silai

Embodied intelligence is moving from demo demonstrations to real-world scenario deployment. The capability boundaries of vision and language models continue to expand, but the shortage of tactile data supply in the physical interaction dimension is becoming the core constraint restricting the improvement of robots' fine operation and scenario generalization capabilities.

At present, the mainstream tactile acquisition solutions in the industry all have their own bottlenecks: film piezoresistive gloves have slip spurious signals and data drift, exoskeleton-style gloves restrict natural grasping postures and introduce noise, and visual-tactile solutions are difficult to achieve full coverage of the palm. Laboratory-grade equipment has excellent performance but is expensive, making it difficult to support mass collection; simulation data can never bridge the reality-virtuality gap.

How to balance lightweight design, full-domain perception, long-term stability and large-scale cost has become a realistic proposition facing the industry.

Recently, flexible fabric sensing enterprise Yaole Technology launched the Tacknit Glove G1 deep tactile data glove, relying on the integrated knitting molding process and full-domain deep tactile perception capability, targeting the large-scale real tactile data collection scenarios for embodied intelligence.

Different from the multi-layer composite stacking sensing idea commonly adopted in the industry, Tacknit Glove G1 adopts fabric flexible sensing technology, which directly mixes and weaves induction yarn and basic yarn, completes the integration of sensing points while the glove body is knitted and formed, eliminating the multi-layer adhesive structure, and fundamentally avoiding the data distortion problem caused by interlayer slip. The weight of a single glove body is about 60g, which minimizes the interference of wearable devices on the natural movement of the human hand and reduces the artificial noise introduced during the collection process.

Schematic diagram of tactile sensing point distribution of Yaole Technology data glove

From the perspective of product capabilities, large-scale data collection scenarios impose multiple constraints on equipment. It is necessary to ensure non-intrusive wearing that does not distort the movement of the human hand; at the same time, full-palm tactile coverage and data stability for long-term collection are required; it is also necessary to take into account continuous operation by multiple personnel, consumable replacement costs, multi-modal data alignment, and reliability for repeated use with high intensity.

Tacknit Glove G1 has 216 effective sensing points built into a single glove, which fully covers the core tactile areas of the fingertips and palm to capture subtle force changes; the glove is washable, which reduces data drift caused by sweat stains and oil stains. With ergonomic versions of multiple sizes, it is not easy to shift when worn for a long time, ensuring the data consistency of long-cycle collection.

Its hardware adopts a split design. The main control box is connected to the glove body through magnetic attraction, and one main control unit can be compatible with multiple pairs of gloves for rotation operation, adapting to multi-person continuous collection tasks; the glove body is replaced as an independent consumable, reducing the marginal cost of large-scale data collection. The wrist main control box integrates an ultra-wide-angle camera to realize multi-modal synchronous collection of tactile sense, hand posture and image; it completes standardized pre-calibration before leaving the factory, and the tactile and image timelines are accurately aligned, opening up the "collection - cleaning - training" link.

Structure of Yaole Technology data glove

Different from most hardware manufacturers that focus on component assembly, Yaole Technology independently develops the whole chain including conductive yarn solutions, customized textile transformation equipment, weaving structures and signal processing algorithms. On the one hand, it iterates the conductive yarn material system independently to ensure long-term signal stability; on the other hand, it transforms textile equipment for the needs of precision sensing, connects the two processes of electronic sensing and traditional weaving, and realizes the simultaneous implementation of flexibility, high precision and mass production.

In terms of commercialization paths, most of the current tactile collection hardware is biased towards highly customized projects, which pushes up the overall acquisition cost and restricts the scale expansion of data sets. Yaole Technology positions the G1 as a standardized industrial consumable, relying on its own production base, reusing the mature quality control system of the automotive sensor production line for tactile gloves, realizing automatic independent production of core yarn preparation and integrated weaving, aiming to lower the threshold for embodied enterprises to obtain tactile data through large-scale mass production. At present, this product has entered the supply chain of leading dexterous hand manufacturers to carry out deployment tests.

In addition to hardware, Yaole Technology plans to build a closed loop of "hardware sales + tactile data service platform". In addition to the delivery of software and hardware kits, it simultaneously outputs value-added services such as raw data cleaning, calibration and classification, and outputs a complete tactile collection solution to the industry.

Yaole Technology data glove

At present, the tactile perception track is in the stage of parallel technical routes. Various solutions such as piezoresistive, capacitive and visual-tactile have their own advantages and disadvantages. A unified industry test standard has not yet been established, and the proportion of customized development is too high, which pushes up the overall deployment cost. The data collection glove with fabric knitting route provides an implementation path, which turns tactile collection from expensive laboratory equipment into a standardized tool that can be deployed in batches.

With the large-scale rollout of low-cost, high-consistency tactile collection terminals, the precipitation of real tactile samples will be accelerated, which will feed back the iteration of models related to dexterous hand force control and object interaction, and promote humanoid robots and dexterous hands to move from laboratory demonstrations to real scenarios such as industrial assembly and domestic services.

In the future, Yaole Technology will continue to iterate fabric tactile perception products, accelerate the expansion of domestic and overseas markets, and cooperate with industry-university-research institutions to participate in the co-construction of tactile collection industry standards, providing independent and controllable flexible tactile perception infrastructure services.