Teleoperation Master-slave Robot Based on Binocular Vision

Abstract

Teleoperation master-slave robots could be used for a variaty of dangerous work, such as live work in power systems. It is important for the operator to get the information on site since the operator usually stay away from the scene. Vision as the most important way for humans to access outside information, is also the first choice for remote operators to perceive the surroundings of the salve robot. In order to let the operator sense the live scene better, we developed a visual system with fisheye binocular vision and ordinary binocular vision, which enables the operator to sense the wide-scale environment around the robot through the fisheye camera and to sense the details of the objects through the ordinary binocular camera. In order to solve the problem of large distortion of the fisheye camera, we focus on the correction and interpolation of the fisheye image, and propose a cosine similar interpolation algorithm, which can better recover the missing pixels of the fisheye image and realize the depth estimation. In addition, combined with our vision device and master-slave robot arm, we designed a teleoperation device with better human-computer interaction performance.