基于单目视觉的机器人避障方法研究

Research on Obstacle Avoidance of Robot Based on Monocular Vision

避障问题一直是机器人领域研究的热点和重点。设计一种基于单目视觉的避障系统,能够同时检测机器人正前方和两侧的障碍物,且对低纹理障碍物同样具有效果。该系统采用特征尺度检测器完成机器人视野正前方障碍物的检测,对于视野两侧的障碍物运用光流平衡策略进行规避,而对于低纹理的障碍物则利用图像熵的方法进行检测。整个系统在地面移动机器人上进行了测试,实验结果表明,该方法能在复杂的环境下可靠地完成机器人的避障任务。

Obstacle avoidance has always been a hot research in robotics field. In this paper,an obstacle avoidance system based on monocular vision is designed,which can simultaneously detect obstacle in front and on both sides of the robot,and has the same effect on low texture obstacle. The system uses the feature scale detector to detect obstacle in the front of the visual field of robot,and uses the optical flow balance strategy to avoid the obstacle on both sides of the visual field,for obstacle with low texture,the image entropy method is applied. The whole system has been tested on ground mobile robot. The experimental results show that the method proposed in this paper can accomplish the obstacle avoidance task reliably in the complicated environment.