Abstract
In recent years the capabilities of mobile ground robots have improved significantly, enabling applications in the field of disaster response for which logistics and efficient transportation of material plays a major role. Mobile robots can assist in these tasks using automated navigation skills by transporting material from A to B. One key ability for safe automated navigation in disaster scenarios is to identify and follow a traversable path using limited a priori information on the environment only. Thus, computing an efficient path which avoids dead ends cannot be guaranteed. A solution to this is to follow a human once through the environment and learn the path on the fly. While person-following approaches for robots already exist, they are typically limited in their functionality and do not take the immediate surrounding and the kinematics of the robot into account. Moreover, existing approaches are usually defined for a specific setup only and cannot be adapted for other systems. In this paper, we present a highly flexible universal person follower for mobile robots which is configurable, not only for the input devices to track a person but also for tracking and path execution. The implementation provides an interface for local planner implementations of the widely used ROS Move Base to follow a tracked path while safely avoiding obstacles within the environment. The person follower is divided into three modules which can easily be adapted to the robot's setup. Experiments in an off-road environment were conducted to test the functionality of the system in a realistic scenario.
| Original language | English |
|---|---|
| Title of host publication | Proceedings of the Austrian Robotics Workshop 2023 |
| Editors | Andreas Müller, Manfred Nader, Hubert Gattringer |
| Publisher | Joanneum Research |
| ISBN (Electronic) | 978-3-200-08957-0 |
| Publication status | Published - 2023 |