Follow a Human using a Mobile Robot Regardless of the Walking Speed

The humanity is ready to merge intelligent robots into their habitual circumstances to work together, share feelings, and make self-reliant compliance. The capability to follow a human autonomously is essential for a mobile robot to interact with humans. Existing work either requires the human to be in line-of-sight or needs a beacon to be installed, which has great constraints on the human mobility. In this paper, we propose an approach to allow a mobile robot to follow a human regardless of his walking speed. In our setup, a human carries a Tango phone that can perform motion tracking using visual inertial sensing and create a map using its RGB-D sensor. The robot localizes itself in the map by incorporating its on-board Kinect and odometry information. The map quality decreases exponentially with increasing human speed. To produce a map that is suitable for the localization and navigation of the mobile robot, we, therefore, propose a map filter algorithm to improve the map quality. Instead of just following the human path, we propose an algorithm (i.e., trajectory filter) to reduce the cost of path and follow the human simultaneously. Experiments are conducted and the capability is illustrated using ROS on a turletbot platform. We also provide a video link to demonstrate our approach at 202. 94. 70. 33/videos/icarm2018. mp4.