AUTHOR=Chen Shouyan , Sun Xinqi , Zhao Zhijia , Xiao Meng , Zou Tao 

TITLE=An online human–robot collaborative grinding state recognition approach based on contact dynamics and LSTM

JOURNAL=Frontiers in Neurorobotics

VOLUME=Volume 16 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2022.971205

DOI=10.3389/fnbot.2022.971205

ISSN=1662-5218

ABSTRACT=Collaborative states recognition is a critical issue for human-robot collaboration during contact task. This paper proposed a flexible contact dynamics and feature selection based state recognition method to identify human-robot collaborative grinding state. The core issue for collaborative grinding states recognition is to distinguish human-robot contact from robot-environment contact. To achieve this, contact dynamic models of both contacts are first constructed to identify the dynamics difference between human-robot contact and robot-environment contact. Considering the reaction speed required by human-robot collaborative states recognition, feature selection based on Spearman correlation and random forest recursive feature elimination are conducted to reduce data redundancy and computation burden. Long short term memory(LSTM) is then used to construct a collaborative states classifier. Experiments results illustrate that the proposed method can achieve a 96% recognition accuracy in a period of 5ms and 99% in a period of 40ms.