FollowMe: A Robust Framework for the Guidance of Sensorless Indoor Mobile Robots

Abstract

In this paper, we present FollowMe, a new system that allows one sensorless robot to autonomously follow another autonomous robot that has multiple sensors. By offloading both the localization and planning computations to the main robot, we are able to maintain a very small hardware requirement on the follower robot (i.e. it only needs to be able to drive). FollowMe works irrespective of what system it is run on and assumes that the main robot has some method of localizing itself and the follower robot, as well as software for navigation (driving itself to a target point). Given this, we can run FollowMe on different kinds of robots, as we have done through our tests on both physical and simulated robots.The FollowMe system is comprised of three main components. The first is the State Machine that takes input from arbitrary localization sources and coordinates through the PathSplitter algorithm to dynamically segment a given path into sequential target positions for each robot. Then, the Evaluator adjusts parameters for both following and path splitting depending on the following performance. As such, FollowMe only handles the queuing of new target points given a predefined path and assumes the master robot can handle the actual driving of each robot (follower and main) to the points. In order to make this possible, the PathSplitter algorithm ensures the follower robot is in view of the main robot at all times so accurate localization can occur. Finally, the state machine has recovery states as needed.After running experiments in both a physical and a simulation environment, we determined that FollowMe is effective at accomplishing the task of guiding both robots along a predefined path accurately, but because of the iterative nature of the following process, it is relatively slow. Our results highlight the potential for using existing autonomous driving technology in robot navigation and suggest promising directions for future research in this area, specifically for use in autonomous wheelchairs or in the warehouse industry.