中文摘要：

为了减少陀螺仪累积误差及外界磁场对机器人航向角的影响,提出了一种基于双目视觉的服务机器人航向角补偿方法。采用颜色分割与重投影相结合的方法,检测图像中的地面区域,通过加权平均融合检测结果,得出较完整的地面图像区域;使用Canny算子提取边缘直线,并进行滤波处理,求出视觉反馈角度,调整机器人的姿态直至反馈角度趋近于零;结合当前陀螺仪方向角和标准方位角,求出此时机器人航向角的偏移角。实验表明,该方法实时性好、航向角精度高,受光照变化的影响较小,完全可以满足室内移动机器人实时导航需求。

英文摘要：

To reduce the effect on the heading angle induced by the accumulated error of the gyroscope and the exter- nal magnetic field.a heading angle compensation algorithm based on binocular vision was proposed. The color seg- mentation method combined with the re-projection transformation was adopted to detect the floor area in the image, and the weighted average method was applied to fuse the detected results to obtain the whole ground plane area. By using the Canny operator, straight edges were extracted and filtered to calculate the feedback angle. Then, the posture of the robot was adjusted until the feedback angle was nearly equal to zero. Therefore,the beading offset angle can be calculated by combining the current angle of the gyroscope with the corresponding standard orientation angle. Experi- mental results indicate that the proposed algorithm, which has good real-time and high accuracy, is robust to illumi- nation changes,and moreover,it can meet the demand of the indoor mobile robot navigation effectively.