DNN and LiDAR Sensor Based Crowd Avoidance Method for Nurse-Following Robot in Healthcare

In a crowded area like a shopping mall, supermarket, airport, railway stations, hospitals, etc. to give the service to a target person we should consider two things. First, we should track the target person in the real environment and then should avoid the crowd or obstacles. In this chapter, we showed that our robot can track a target nurse using a color histogram matching algorithm and can follow that target nurse. We also showed that if the nurse is occluded by pedestrian persons or a crowd our robot can avoid that pedestrian person or crowd and again track that target nurse. We use an omnidirectional camera and DNN-based human pose estimator called OpenPose to locate the position of the target nurse and pedestrian persons. Then, we use the LiDAR sensor to calculate the distance. For crowd avoidance, we use the line following and circle following method. Assuming that the robot was blocked from the target nurse in various situations and all the situations our robot successfully avoids the crowd. Our result confirms that the robot can avoid the crowd successfully.