基于单目视觉的移动机器人跟随

Person following of mobile robot using monocular camera

针对室外环境下移动机器人基于单目视觉对目标人实时跟随问题进行了研究,提出通过均值漂移算法对像平面目标的跟踪来实现目标定位。为了补偿摄像机运动造成目标在像平面的偏移,提出了以卡尔曼滤波器的估计值作为均值漂移算法的启动点,然后利用均值漂移算法获得最终位置;为了估计单目视觉下目标与机器人之间的距离,提出了一种借助于主颜色描述子和形状直方图的间接景深计算方法。室外实验表明该算法具有很好的实时性和鲁棒性,能有效实现复杂室外环境下对人的跟随。

The method for person following of mobile robot using monocular camera in outdoor environment is researched, the location of object is achieved through using the mean shift algorithm,which realizes object tracking in the image plane.To compensate the offset of the target appearing in the image sequences caused by the motion of the camera,an algorithm that combines the mean shift algorithm and Kalman filter is proposed,with the state estimation of filtering as the starting position of the mean shift and with the converge location of the mean shift as the tracking results in current frame.In order to estimate the distance between the object and robot,an indirect depth computation approach that adopts dominant color descriptor and shape histogram is proposed.Experiment results indicate that the proposed algorithm is robust,and has good tracking performance and real-time characteristic on person following under complex outdoor environment.