Research On Omni-directional Vision Navigation Of Mobile Robot

Since the first robot was produced in the later 1950s, robots have been widely used in varieties of industries. Mobile robot navigation based on computer vision is one of the hot topics in robot field recently. And in the behaviors of mobile robot navigation, the object tracking and the obstacle avoidance are the two most basic tasks. The Pioneer3 mobile robot produced by ActivMedia Robotics Corporation is taken as research platform to design and implement object tracking and obstacle avoidance system.Firstly, the history and the status of computer vision are introduced. The status and challenges of mobile robot vision navigation are discussed in detail.Secondly, the basic approaches of the color image processing are presented, which include the choice of color model, image preprocessing and binary morphology operation. The transformation between the different color space and the improved conversion formula are introduced in detail.Finally, the object tracking system and the obstacle avoidance system are implemented. Approach based on color look-up table is used to improve real-time character of the object tracking system. At the same time, in order to complete object recognition, the methods of drawing object character parameters is introduced. After that, a fast system for avoiding unknown obstacles on a mobile robot using a panoramic color camera as the only sensor is introduced. Two study modes are used to adapt to the change of environment. In the manual mode, the system is easily trained by simply driving the robot through its environment. In the autonomous mode, the system keeps learning the color of the ground during operation. The algorithm performs in real-time and provides a high-resolution ground-obstacle binary image for robot decision-making system.