Real-Time Obstacle Avoidance by Visually Recognizing Laser Patterns

Abstract

This paper presents a 3-D local map building approach for real-time obstacle avoidance using visually-recognized laser patterns. Precise estimate of the local map provides essential information for navigation and control of a mobile robot in unknown environments. Existing navigation and control approaches typically require expensive sensing devices, such as sonar or laser range finder. In this paper, a mobile robot mounted with a CCD camera and a laser line projector is proposed for the considered control tasks. The idea is to reconstruct the actively-projected laser line in Cartesian space from its observed image based on known geometrical relation between the laser line projector and the CCD camera. The position of the obstacle can thus be estimated based on the reconstructed laser line for effective and efficient navigation of mobile robot. The proposed system has been effectively validated in laboratory environments by performing experiments with a custom-made wheeled robot.