Person Following Mobile Robot using Pedestrian Dead-Reckoning with Inertial data of Smartphones

Abstract

For a person following mobile robot, identifying the specific host to follow is challenging in presence of other people with similar features in its camera frame or field of view. In this paper we propose the use of Pedestrian Dead-Reckoning using the Inertial Measurement Unit (IMU) data from smartphone of host or user, which is used to localize the user in the map frame of the mobile robot. The movement and orientation of the user is used to calculate the path of the user and used as global targets for the mobile robot to follow. The mobile robot which is doing Simultaneous Localization and Mapping (SLAM), running on the Robot Operating System (ROS) framework follows the user as the user moves while it also implements obstacle avoidance locally. Since the smartphone is uniquely paired with the mobile robot, the ambiguity of identifying the host is also solved.