Development of the Bioinspired Tendon-Driven DexHand 021 with Proprioceptive Compliance Control
Jianbo Yuan, Zhu Haohua, Jing Dai, Sheng Yi
AI summary
Problem
Replicating the human hand’s dexterity, compliance, and sensing in a lightweight, industrial-grade robotic system remains challenging due to excessive weight, control inefficiencies, and the difficulty of miniaturizing force sensors in compact mechanisms.
Approach
The authors developed a 1 kg, 19-DOF tendon-driven robotic hand with an artificial muscle design and implemented a Gaussian Process Regression-based joint torque estimation model to enable proprioceptive admittance control without external force sensors.
Key results
- 1 kg weight with 19 DOFs (12 active) in a compact form factor
- Single-finger load capacity exceeding 10 N and fingertip repeatability under 0.001 m
- 31.19% reduction in multi-object grasping joint torques compared to PID control
- Force estimation errors below 0.2 N with stable thermal management under 70°C
Why it matters
Provides a scalable, lightweight, and compliant manipulation platform for industrial automation and intelligent manufacturing where safety, precision, and cost-effectiveness are critical.
Abstract
The human hand plays a vital role in daily life and industrial applications, yet replicating its multifunctional capabilities-including motion, sensing, and coordinated manip- ulation with robotic systems remains a formidable challenge. Developing a dexterous robotic hand requires balancing human- like agility with engineering constraints such as complexity, size- to-weight ratio, durability, and forcesensing performance. This letter presents DexHand 021, a high-performance, cable-driven five-finger robotic hand with 12 active and 7 passive degrees of freedom (DOFs), achieving 19 DOFs dexterity in a lightweight 1 kg design. We propose a proprioceptive force-sensing-based admittance control method to enhance manipulation. Experi- mental results demonstrate its superior performance: a single- finger load capacity exceeding 10 N, fingertip repeatability under 0.001 m, and force estimation errors below 0.2 N. Compared to PID control, joint torques in multi-object grasping are reduced by 31.19 %, significantly improves force-sensing capability while preventing overload during collisions. The hand excels in both power and precision grasps, successfully executing 33 GRASP taxonomy motions and complex manipulation tasks. This work advances the design of lightweight, industrialgrade dexterous hands and enhances proprioceptive control, contributing to robotic manipulation and intelligent manufacturing.