Spatiotemporal mapping and similarity evaluation based on isomer motion

In the process of human-robot interaction, some good effects can be achieved with the help of motion similarity assessment when the demonstrator and the imitator have similar configurations. However, the motion similarity assessment method is ineffective in the isomeric situation. As the robot application range extends constantly, there is a growing need for action evaluation and demonstration of teaching among isomers. This paper presents a spatio-temporal mapping and similarity assessment method based on the movements of isomer. For the problem of motion time sequence alignment and spatial mapping between isomers, the key frames of motion sequence are obtained by improved k -means clustering method and reconstruction of joint matrix. Based on the aligned motion sequences, the Kabsch algorithm is used to calculate dynamic metric. In addition, there is a static metric to represent the shape of the isomer. Finally, the two metrics are combined by the similarity comparison algorithm to obtain the motion similarity percentage. In the comparison of movement similarity between human and robotic arm, the average percentage of similarity obtained by this method was more than 75.06% for the same movement, indicating that the movement similarity between isomers could be correctly identified, and the average performance was improved by 23.46% compared to the BPE evaluation model.