一种无监督护理机器人遥操作员训练与评估方法  

Unsupervised Training-evaluation Method for Nursing Telerobot Operator

作　　者：朱德良 谢劼欣 严海涛 贾路宽 郭士杰 ZHU De-liang;XIE Jie-xin;YAN Hai-tao;JIA Lu-kuan;GUO Shi-jie(College of Mechanical Engineering,Hebei University of Technology,Tianjin 300130,China;Academy for Engineering and Technology,Fudan University,Shanghai 200433,China)

机构地区：[1]河北工业大学机械工程学院,天津300130 [2]复旦大学工程与应用技术研究院,上海200433

出　　处：《小型微型计算机系统》2023年第12期2640-2645,共6页Journal of Chinese Computer Systems

基　　金：上海市科委“科技创新行动计划”项目(21511101701)资助;国家自然科学基金项目(61871173)资助;河北省重点研发计划项目(19211817D)资助。

摘　　要：针对遥操作护理机器人操作员培训难度大,训练效率低的问题,本研究提出了一种基于无监督轨迹分割的操作员培训方法.该方法无需专家参与,通过对比划分的子轨迹与标准任务流程轨迹的差异,为操作员提供更加直观和程序性的评估方案.为了实现这一目标,本研究构建了遥操作护理机器人系统,并对机器人轨迹从粒度上进行建模,为遥操作护理机器人操作员的培训提供可靠的反馈标准.同时提出了一个新型的TSC-CRP无监督聚类模型,该模型可以在无需人工标注的情况下对机器人轨迹进行无监督轨迹分割.实验证明,TSC-CRP的分割精度较其他无监督方法提高了32.2%~38.1%,并可以识别不同操作员的技能水平.因此,本文提出的方法可以有效的提高护理机器人操作员的训练效率,并对操作员的技能水平进行全面客观的评估.To cope with the difficulty of training and evaluation for nursing telerobot operator.This paper proposes a training-evaluation method for operator with unsupervised trajectory segmentation.To evaluate the dexterity and procedural knowledge of the operators objectively,we propose a new unsupervised model TSC-CRP that can automatically segment trajectory from nursing robotic training sessions.By comparing the segmented sub-trajectories and the standard sub-trajectory process,the method can provide objective evaluation and meaningful feedback without the intervention from experts.Experiments show that TSC-CRP has higher segmentation accuracy than other unsupervised methods with 32.2%~38.1%,and it can identify the operators with different skill levels.Therefore,the proposed training-evaluation method can effectively improve the training efficiency of nursing robot operators,assess their skills objectively.

关 键 词：遥操作机器人 技能评估 无监督聚类 轨迹分割 

分 类 号：TG156[金属学及工艺—热处理]