3D visual sensing technique based on focal stack for snake robotic applications

Abstract

In this paper, we proposed a passive 3D visual sensing technique based on the focal stack and designed the snake robot’s 3D visual sensing prototype. The proposed technique consists of depth reconstruction, all-in-focus imaging and 3D point cloud generation based on the theoretical model of light field reconstruction from the focal stack. The 3D visual sensing prototype was realized by the focal stack acquisition system controlled by the servo motor. In the experiment, the depth map with high precision, all-in-focus image with high spatial resolution as well as the 3D point cloud can be reconstructed in the field-of-view (FOV) by the 3D visual sensing prototype. The proposed technique could be further applied to efficient visual navigation, simultaneous-localization-and-mapping (SLAM) and visual servoing of the robot. Keywords: Visual sensing, Focal stack, Light field imaging, 3D imaging, Snake robot