Automating a Telepresence Robot for Human Detection, Tracking, and Following

The operation of a telepresence robot as a service robot has gained wide attention in robotics. The recent COVID-19 pandemic has boosted its use for medical uses, allowing patients to interact while avoiding the risk of contagion. While telepresence robots are designed to have a human operator that controls them, their sensing and actuation abilities can be used to achieve higher levels of autonomy. One desirable ability, which takes advantage of the mobility of a telepresence robot, is to recognize people and the space in which they operate. With the ultimate objective to assist individuals in office spaces, we propose an approach for rendering a telepresence robot autonomous with real-time, indoor human detection and pose classification, with consequent chaperoning of the human. We validate the approach through a series of experiments involving an Ohmni Telepresence Robot, using a standard camera for vision and an additional Lidar sensor. The evaluation of the robot's performance and comparison with the state of the art shows promise of the feasibility of using such robots as office assistants.