Navigation strategy for the service robot in the elevator environment

Recently, various service robot systems are introduced and offer the guidance and delivery service. However, most of them are operated on a single floor, because they do not consider the elevator environment. In this paper, we present the navigation strategy for multi-floor environment including elevator situation. Firstly, we propose the recognition method of the button and the floor number so that the robot reacts flexibly to a current elevator status. We apply adaptive threshold method to a current image in order to get a binary image and make the candidates of the button and the floor number using additional filtering method. By rejecting weak candidates using the artificial neural network, we finally get the position of button and current floor number. Secondly, we suggest the path planning algorithm for navigating in the elevator. By constructing an occupancy grid map of the inner part of the elevator, we calculate the optimal position to move and make a shortest path using the distance transform.