Towards Socially Acceptable, Human-Aware Robot Navigation

Abstract

The introduction of service robots to our daily life requires adaptation of the current navigation strategies. In the presence of humans, robots must be designed to ensure their safety and comfort. This paper proposes a layered costmap architecture that incorporates social norms to generate trajectories compatible with human preferences. The implemented framework creates a social abstraction of the environment – in the form of an occupancy grid – to plan human-friendly paths. It employs information about individuals in the scene to model their personal spaces. In addition, it uses predicted human trajectories to improve the efficiency and legibility of the robot trajectory. Different simulation scenarios resembling everyday situations have been used to evaluate the proposed framework. The results of the experiments have demonstrated its ability to behave according to social conventions. Furthermore, the navigation system was assessed in real life experiments where it was proved capable of following similar paths to those performed by humans.