面向复杂力交互任务的操作技能传递与控制研究  

Research on Operation Skill Transfer and Control Oriented to Complex Force Interaction Tasks

作　　者：赵杰[1] 武睿[1] 张赫[1] 朱延河[1] 臧希喆[1] ZHAO Jie;WU Rui;ZHANG He;ZHU Yanhe;ZANG Xizhe(State Key Laboratory of Robotics and System,Harbin Institute of Technology,Harbin 150001)

机构地区：[1]哈尔滨工业大学机器人技术与系统国家重点实验室,哈尔滨150001

出　　处：《机械工程学报》2022年第18期116-132,共17页Journal of Mechanical Engineering

基　　金：国家自然科学基金集成“面向航空制造的人-机器人协作技术及应用研究”(92048301)资助项目。

摘　　要：如何让机器人拥有像人一样强大的感知能力并执行复杂操作,尤其是带有力交互的复杂操作是机器人学界一直探索的问题。这个问题的解决,能够帮助机器人实现从“设备”向“助手”的转化。而面向复杂力交互任务的操作技能传递与控制作为当前人-机技能传递领域研究的前沿方向之一,其研究核心是实现对熟练操作者力交互操作过程中的多模态技能数据进行示教学习,并通过设计合理的技能模型,结合先进的控制理论以及机器人感知能力,实现机器人自主执行复杂力交互任务的目的,从而让机器人真正的可以协助甚至代替人类执行生活中常见的复杂任务。总结该领域较为重要的三个问题:(1)多模态信息融合的示教方式;(2)针对力交互任务的技能学习;(3)基于机器人柔顺控制的技能控制与基于机器人感知的智能技能切换;并对该领域的研究现状展开分析和讨论。The problem that has been explored for years in robotics is how to make robots have the same strong perception capabilities as humans and perform complex operations, especially with force interaction. Solving the problem can help the robot to truly transfer from “device” to “assistant”. The transfer and control of operational skills for complex force interaction tasks is one of the frontiers of the human-robot skills transfer research field. The core of this research lies in the learning of multi-modal skill data, demonstrated in the force interaction operations process by skilled operators. And through reasonable design skill models, combined with advanced control theory and robot perception capabilities, the robot can autonomously perform complex force interaction tasks, so that robots can truly assist or even replace humans in performing common complex tasks in life. This article summarizes three important issues in this field:(1) Multimodal information fusion demonstration method;(2) Skill learning for force interaction tasks;(3) Skill control based on robot compliance control and intelligent skill switching based on robot perception. At last, the research status in this field will be discussed and analyzed.

关 键 词：技能传递 变阻抗控制 多模态信息融合 力交互 

分 类 号：TG242[金属学及工艺—铸造]