Zhongke Huisi Unveils Three Dexterous Hands, Targeting Practical Task Execution for Embodied Robots

Zhongke Huisi Introduces a Family of Dexterous Hands in Changsha

Zhongke Huisi, a venture formed by robotics firm Zhongke Huiling, manufacturer Lens Technology, and Hunan Huaxia Group, publicly launched three dexterous hand products on 7 August in Changsha's Xiangjiang New Area. The products—L1, D1, and M1—are joined by a pre-announced F-series tendon-driven hand, marking a broad push to move embodied artificial intelligence from basic locomotion to fine manipulation.

The company traces its technology back to multidisciplinary research at the Institute of Automation, Chinese Academy of Sciences (CASIA), which had been working on humanoid robot brains, sensing, and interaction before 2023. Zhongke Huiling was founded to productize those results, and later the team decided the dexterous hand required a dedicated entity. Chairman Zhang Zhengtao described the goal simply: "We hope to make the 'dexterous hand' a hand that can actually do work."

The market backdrop is rapidly expanding. According to a GGII report for H1 2026, China's dexterous hand market sales reached about 19,200 units in 2025, up 236.84% year-on-year, and are expected to hit 70,200 units in 2026. From 2023 to May 2026, roughly 128 billion yuan was raised across 57 financing rounds in the sector.

The Strategy Behind Splitting Hands by Use Case, Not Just Specs

Segmentation by Task, Not Just by Degrees of Freedom

L1 is a lightweight (610g), low-cost option with a grasp load of 25 kg and a full-hand open/close time of just 0.2 seconds, priced in the thousands of yuan. It has already been proven in musical instrument playing by the CASBOT BAND. D1 is a high-degree-of-freedom five-finger hand with 16 active and 4 passive DOFs, equipped with piezoresistive tactile sensors at the fingertips. M1 adopts a modular architecture with a standardized body and swappable functional fingertips. The pre-released F-series explores a fully tendon-driven design reaching 37 DOFs and claims a key maintenance breakthrough: tendons can be replaced in three minutes, tackling the long-standing problem of tendon degradation.

Moving the Industry Benchmark from Specs to Real Tasks

Zhang explicitly criticized the industry's obsession with parameter comparisons—DOF counts, load, precision—and argued that real-world task completion should be the true yardstick. The company is trying to break the deadlock where high-DOF hands cost hundreds of thousands of yuan, face severe thermal and integration challenges, and suffer from rapid accuracy decay. Their layered response separates "can it work" (L1's reliability and cost), data collection (D1's sensing), task-switching efficiency (M1's modularity), and maintainability (F-series' fast tendon swap).

A Data Flywheel Anchored in Hardware Diversity

Zhongke Huisi is coupling its product matrix with what it calls a "multi-body × data acquisition" dual flywheel. The varied hand designs serve as stable platforms for gathering real-world manipulation data. CEO Yang Ping noted the industry's emerging data ratio consensus—70% synthetic/simulation, 20% real human demonstration, 10% real robot data—and the company plans to build what it claims will be China's largest dexterous hand skill training facility in Xiangjiang New Area, with a 1,200-square-meter site already allocated. The partnership structure embeds this: CASIA provides core robotics IP, Lens Technology offers precision manufacturing and supply chain muscle, and Huaxia Group contributes tourism, education, and industrial application scenarios.

Implications for Robot Builders and the Dexterous Hand Supply Chain

For robotics integrators and original equipment manufacturers, the product range creates clear procurement options. L1 addresses high-reliability, cost-sensitive applications where a rapid open/close cycle and robust grip matter more than high dexterity. D1 is suited for research labs and data-hungry reinforcement learning pipelines that need tactile feedback. M1's interchangeable fingers mean operations that switch between delicate and forceful tasks on the same robot may find a fit.

For suppliers of micro-actuators, sensors and tendons, the F-series' claim of a three-minute tendon change, if realized at scale, could shift demand toward modular, user-serviceable tendon units rather than sealed assemblies. Suppliers should watch for specification requirements from the Xiangjiang training facility.

For investors monitoring the embodied AI supply chain, the 236% YoY unit growth and anticipated 43,000-unit market by 2030 signal that the bottleneck is moving from "can it walk" to "can it manipulate." Companies with practical task validation—like musical instrument playing here—may differentiate themselves from pure-spec competitors. However, the note from Yang that "the hand itself is not the core; the complete operation, interaction and motion control capability is the moat" implies that pure-play hand makers without a system integration angle may face commoditization pressure.