Multimodal Navigation for Autonomous Service Robots

摘要

This paper introduces a navigation stack for service robots with different modalities, namely human-in-the-loop navigation with people following or guiding modes, as well as standard waypoint navigation. The work also focuses on dependable localization and obstacle avoidance, since these are important components in a fully functional navigation stack. The proposed approach is based mostly in state of the art algorithms, integrated in a service robot to offer safe and dependable navigation modalities. A novel method to follow a dynamic goal is proposed, enabling the robot to follow goals whose position and orientation may move with time, whether it is a person or a moving object. The method was implemented in a real robot and validated through experiments in the ISRoboNet@Home testbed (domestic environment) and in the competition European Robotics League Smart Cities (coffee shop environment).