人机交互下遥操作机器人的精细化力感知与力控制技术研究

Compared with the traditional teleoperation technology, the precise teleoperation has strict requirements on the perceptual accuracy of the interaction force, the accurate recognition of the dynamic stiffness of the operation object, and the processing of the system delay. In response to this problem, this project intends to carry out basic theoretical research on the precise force perception and force control technology of teleoperation robots under human-computer interaction. Firstly, based on multi-wavelength fiber grating sensing, a multi-dimensional small force tactile sensing model with temperature compensation is proposed. The static decoupling algorithm based on coupling error modeling is used to decouple multi-dimensional interaction information to obtain multi-dimensional precise force tactile information. Then anti-interference stable grabbing control method based on dynamic stiffness sensing is proposed. The gripping force is dynamically adjusted during the grabbing process according to the dynamic interaction force tactile information, and the output force is precisely controlled by the force control law in the Cartesian Space to achieve anti-interference and stable grabbing. Finally, for the delay problem under the precision control of teleoperation, the system delay is compensated by the motion prediction in the virtual interaction model, and then the control strategy for tracking error between the bilateral position of the master-slave robot is established, and the non-time reference based control system is used to avoid the influence of delay, and the precision force control of the teleoperation robot under human-computer interaction is realized.

与传统遥操作控制相比，精细化作业下的遥操作控制对末端交互力触觉的感知精度，被操作对象的动态刚度精准识别，系统时延的处理等方面有着苛刻要求。针对目前遥操作机器人的力感知精度及力控制稳定性无法满足精细化作业需求的问题，本项目拟围绕人机交互下遥操作机器人的精细化力感知与力控制技术展开基础理论研究。首先，基于多波长光纤光栅传感提出具有温度补偿的多维微小力感知模型，利用基于耦合误差建模的静态解耦算法对微小力交互信息进行多维解耦，以获取多维精细化力触觉信息。然后提出基于动态刚度感知的力封闭抗干扰稳定抓取控制方法，在抓取过程中根据动态交互力触觉信息动态调整抓取力大小，利用笛卡尔空间力控制律对输出力进行精密控制，以实现抗干扰稳定抓取。最后针对遥操作精密控制下的时延问题，通过虚拟交互模型中的运动预测对系统时延进行补偿，然后利用基于非时间参考的控制系统避开时延的影响，实现人机交互下遥操作机器人的精密力控制。

与传统遥操作控制相比，精细化作业下的遥操作控制对末端交互力触觉的感知精度、被操作对象的动态刚度精准识别、系统时延的处理等方面有着苛刻要求。针对目前遥操作机器人的力感知精度及力控制稳定性无法满足精细化作业需求的问题，本项目围绕人机交互下遥操作机器人的精细化力感知与力控制技术展开了系统研究，在IEEE TII、SCIS、IEEE TMech、IEEE TIM、中国科学:信息科学等国内外高水平期刊和国际会议发表学术论文20篇，授权发明专利3项，超额完成了既定研究任务与目标。主要研究成果包括：1）以FBG作为传感载体构建了可测量多维微小力应变且具有温度补偿效果的力感知通用模型，建立了中心波长与多维微小力应变的函数关系，提出多维力维间解耦方法，达到了克服外界环境变化影响和提高多维微小力应变测量精度的目的。2) 基于机器人抓取过程中，根据机器人力触觉传感器反馈的动态高精度接触力触觉信息，实现了机器人多模态感知，提升了机器人的感知能力，解决了机器人抓取时环境干扰下抓取力的精确力控制问题。3）提出人机交互下遥操作控制系统的主从机器人力分配机制，在保证遥操作系统稳定性前提下，提高系统的透明度，构建了力分配的机制，分析了主从机器人力传递的关系，并通过数值模拟和虚拟物理模拟分析了接触力转换机制。