基于多智能体思想的辅助手部运动系统设计

Design of a hand-assiting system based on a multi-agent model

考虑运用多智能体思想,辅助手部活动不便患者实现日常的抓握、伸直等动作。于Unity引擎内设计并实现了一个基于多智能体蜂拥算法的辅助手部运动系统的仿真模拟,并对基础蜂拥算法进行了改进,使其更符合手指的运动方式。该系统包含80个智能体(智能体安装在类似手套的载体上,且与载体之间由弹簧关节连接)以及5个可随意变换位置的虚拟目标点,5个目标点的位置分别代表患者左手5根手指意图到达的位置。辅助系统启动后,蜂拥算法控制的智能体开始自组织振动,通过弹簧带动手指逐渐运动至目标点,最终实现患者的运动意图。通过调整各目标点的位置,可充分辅助患者实现基本的抓握动作。该多智能体系统可根据实际情况灵活改变智能体的数目和大小,且具有良好的稳定性,为多智能体系统的应用提供了崭新的思路。

Multi-agent system(MAS)have many uses in fields such as flight formation and signal control.This paper investigates the use of MAS in rehabilitation medicine to help patients with hand defects to carry out daily activities.A simulation of the hand-assisting system based on the flocking algorithm was designed and implemented in the Unity engine.The basic flocking algorithm was improved to make it more suitable for finger movement.This hand-assisting system consists of 80 agents which are mounted on the carrier such as a glove with a spring joint between the carrier and the finger,and five virtual target points that can be freely changed.The positions of the five target points represent the locations where the five fingers are intended to arrive.After turning on the hand-assisting system,the agent begins self-organized vibration,following by gradual movement of the finger towards the target point through the spring joint,finally realizing the patient s intended movement.By adjusting the position of each target point,it is possible to fully assist the patient in achieving a basic grasping action.The MAS can change the number and size of the agents flexibly and has good stability.This work suggests a new application for MAS.