Behavior classification and image processing for biorobot-rat interaction

Abstract

In this paper, we focus on rat's behavior classification for biorobot-rat interaction. The automatic behavior analysis and classification of laboratory rats can effectively improve the adaptivity of interaction between rat-like robot and biological rats. Basic image processing algorithm as Labeling and Contour Finding were employed to extract feature parameters (body length, body area, body radius, rotational angle, and ellipticity) of rat's actions. These feature parameters are integrated as the input feature vector of CNN (Convolutional Neural Network) and SVM (Support Vector Machine) training system respectively. Preliminary experiment result shows that the grooming, rotating, crouching and rearing actions could be recognized with extremely high rate (more than 90%) by both CNN and SVM. Furthermore, CNN provides better recognition rate and SVM provides less computational cost.