Pose estimation of small-articulated animals using multiple view images

Abstract

The study of robots that mimic the behavior of small animals such as lizards and arthropods has been actively carried out. However, systematic analysis of the gait behavior of small animals is not prevalent in the literature. Motion analysis of most living creatures is conducted using the optical motion capture system. However, it is inapplicable to small animals owing to the difficulty of attaching optical markers. To solve this problem, markerless motion capture is being researched; however, many of these studies were performed on human subjects. Therefore, to apply a markerless motion capture system for insects that have many legs and a high degree of freedom, additional research is required. The objective of this study is to develop a system to estimate the continuous pose of small-articulated animals using a three-dimensional skeleton model of the animal from a multi-view video sequence. It includes the extraction of the extremity and root of each leg and the calculation of joint kinematics using the forward and backward reaching inverse kinematics (FABRIK) algorithm that uses the extracted extremity and root. The method developed in this study will contribute to better understand the gait behavior of small-articulated animals.