Autonomous Person Following for Telepresence Robots Akansel Cosgun, Dinei A. Florencio and Henrik I. Christensen Abstract— We present a method for a mobile robot to follow a person autonomously where there is an interaction between the robot and human during following. The planner takes into account the predicted trajectory of the human and searches future trajectories of the robot for the path with the highest utility. Contrary to traditional motion planning, instead of determining goal points close to the person, we introduce a task dependent goal function which provides a map of desirable areas for the robot to be at, with respect to the person. The planning framework is flexible and allows encoding of different social situations with the help of the goal function. We implemented our approach on a telepresence robot and conducted a controlled user study to evaluate the experiences of the users on the remote end of the telepresence robot. The user study compares manual teleoperation to our autonomous method for following a person while having a conversation. By designing a behavior specific to a flat screen telepresence robot, we show that the person following behavior is perceived as safe and socially acceptable by remote users. All 10 participants preferred our autonomous following method over manual teleoperation. I. I NTRODUCTION Mobile telepresence robots constitute a promising area for the robotics industry, and early systems are already commer- cially available from companies such as Vgo Communica- tions, Anybots, and Double Robotics. These are small scale deployments, and a number of issues need to be addressed before wide deployment, many posing interesting technical challenges. These challenges include designing more intu- itive user interfaces [1, 2], better video conferencing [3, 4], better audio capture [5, 6], how the remote user’s presence is displayed [7], overcoming wireless connection restrictions [8, 9], assisting teleoperation [10, 11] and adjusting the level of autonomy [11, 12]. Telepresence robots are a level above video conferencing since the robot is used as the communication medium and the remote user can now control the movement. Therefore, the spatial interaction between people and a telepresence robot in social situations is worth investigating. One of those situations is moving with a group of people. In an effort to analyze the spatial and verbal interaction, we focus on engagement with one person where the remote user interacts with the person while following him/her in a corridor. This is situation is very likely to happen in office environments, for example when the remote user is having a discussion with a co-worker while walking to his office after a meeting. As telepresence robots become more common, there will be need to have the functionality A. Cosgun is a Ph.D. student at Georgia Tech, Atlanta, GA. H. Christensen is the director of Center for Robotics and Intelligent Machines at Georgia Tech, Atlanta, GA. D. Florencio is a researcher with Microsoft Research, Redmond, WA. Fig. 1: Telepresence robot user and a present person interacting while walking together. of autonomous following of a person so that the remote user doesn’t have to worry about controlling the robot. In this paper, we propose a planning framework for au- tonomous person following. As opposed to traditional motion planning, our approach does not determine explicit goal states with respect to the person but achieves the desired behavior by searching for the best utility over time. The framework considers the future trajectory of the followed person so that we can account for moving targets. The robot behavior can be adjusted with the help of the goal and cost functions defined for a specific task. The planned path is dependent on the location of the person in relation to the robot, distance to obstacles as well as the velocity and acceleration of the robot. We evaluate our system by first showing quantitative results for the performance and then conducting a user study. The user study aims to compare manual and autonomous person following when the remote user has a task at hand that involves interaction. The task consists of listening to a passage the followed person reads and answering related questions afterwards. We also observe subjects’ experiences using the system, get useful feedback and pinpoint future challenges that can be helpful designing new applications for telepresence robots. First we examine the relevant literature in Section II in- cluding person following, social navigation and telepresence. The robot platform used in this work is described in Section III. Section IV describes our algorithm in detail. Section V presents our evaluation and the user study and we discuss the results in Section VI.