基于卷积神经网络的服务机器人人体跟踪算法

Human Body Tracking Algorithm of Service Robot Based on Faster-RCNN

人体跟随是服务机器人在人机交互领域的研究热点。当存在障碍物对人体进行遮挡时,单一算法的服务机器人人体跟随效果较差。提出一种Faster-RCNN与卡尔曼滤波相结合的算法,使用Faster-RCNN检测图像中的人体目标,建立相应的卡尔曼滤波器进行位置估计,使用跟踪过程中Faster-RCNN检测到的人体位置信息来更新卡尔曼滤波器。实验验证了在有障碍物遮挡的情况下,人体跟随效果得到提升。

Human body following is a research hotspot of service robots in the field of human-robot interaction.When the human body covered by obstacles,the effect of the service robot following with a single algorithm is poor.An algorithm combining Faster-RCNN and Kalman filtering is proposed in this paper.Faster-RCNN is used to detect human targets in the image.Corresponding Kalman filters are established for position estimation.The position information of the human body detected by Faster-RCNN is used to update the Kalman filter during tracking.Experiments verify that the tracking effect is improved when obstacles cover the human body.