Obstacle avoidance of autonomous mobile robots

Abstract

Robots have been widely used in industry to replace human being for repeated and dangerous works. Typically, the robot is used to pick and place workpieces or to transport material in a way of unmanned vehicle. This gives rise to the issue of obstacle avoidance. There are model-based and sensor-based methods for this purpose. In a model-based system, one uses artificial intelligence search algorithms to find free path in the working environment. In a sensor-based system, the sensed data is recorded in a grid map. Obstacles are represented by cells in the map. The value in the cell determines whether there exists an obstacle or not. We use the ultrasonic sensors to detect obstacles. The sensors are installed every thirty degrees at the center of robot. The surrounding information can be obtained at once. However, the poor range capability of the sensor and the shape of the obstacle may influence the performance of obstacle avoidance, we also use two CCD cameras to identify and locate target, obstacles. The pattern recognition is essentially done by the calculation. Of the first and second moment of the image. The obstacle is located by the comparison of the above calculations from two cameras. The detailed methods and experimental results will be presented with video tape.