# An Adaptive Grasping Multi-Degree-of-Freedom Prosthetic Hand with a Rigid–Flexible Coupling Structure

## TL;DR

This paper introduces a prosthetic hand with a flexible structure and control system that can grasp objects more adaptively.

## Contribution

A novel rigid-flexible coupling structure and TCN-based sEMG-to-force mapping for adaptive grasping in prosthetic hands.

## Key findings

- The rigid-flexible coupling structure provides an adaptive joint without additional actuators.
- The prosthetic hand successfully grasps flexible objects of various shapes and sizes.
- Improved TCN mapping enhances accuracy in converting sEMG signals to grasping force.

## Abstract

What are the main findings?
A multi-degree-of-freedom prosthetic hand with a rigid–flexible coupling structure.Control algorithms for flexible object grasping.Mapping surface electromyographic (sEMG) signals to force using temporal convolutional network (TCN).

A multi-degree-of-freedom prosthetic hand with a rigid–flexible coupling structure.

Control algorithms for flexible object grasping.

Mapping surface electromyographic (sEMG) signals to force using temporal convolutional network (TCN).

What is the implication of the main finding?
It can provide an additional adaptive joint for the finger without adding actuators.It is more suitable for grasping flexible objects than traditional algorithms.It provides a more accurate method to map sEMG signals of eight channels to force signals.

It can provide an additional adaptive joint for the finger without adding actuators.

It is more suitable for grasping flexible objects than traditional algorithms.

It provides a more accurate method to map sEMG signals of eight channels to force signals.

This study presents the design and evaluation of a dexterous prosthetic hand featuring five fingers, ten independently actuated joints, and four passively driven joints. The hand’s dexterity is enabled by a novel rigid–flexible coupled finger mechanism that incorporates a 1-active–1-passive joint configuration, which can enhance the dexterity of traditional rigid actuators while achieving a human-like workspace. Each finger is designed with a specific degree of rotational freedom to mimic natural opening and closing motions. This study also elaborates on the mapping of eight-channel electromyography to finger grasping force through improved TCN, as well as the control algorithm for grasping flexible objects. A functional prototype of the prosthetic hand was fabricated, and a series of experiments involving adaptive grasping and handheld manipulation tasks were conducted to validate the effectiveness of the proposed mechanical structure and control strategy. The results demonstrate that the hand can stably grasp flexible objects of various shapes and sizes. This work provides a practical solution for prosthetic hand design, offering promising potential for developing lightweight, dexterous, and highly anthropomorphic robotic hands suitable for real-world applications.