Enhancing Navigational Safety in Crowded Environments using Semantic-Deep-Reinforcement-Learning-based Navigation

TL;DR

This paper introduces a semantic deep reinforcement learning method for mobile robot navigation in crowded environments, improving safety by considering object-specific danger zones and adaptive safety distances.

Contribution

It presents a novel semantic DRL approach that incorporates high-level obstacle information and object-specific safety rules for safer navigation in dynamic crowds.

Findings

Increased safety compared to benchmark obstacle avoidance methods.

Ability to learn object-specific safety distances.

Enhanced navigation safety in highly dynamic environments.

Abstract

Intelligent navigation among social crowds is an essential aspect of mobile robotics for applications such as delivery, health care, or assistance. Deep Reinforcement Learning emerged as an alternative planning method to conservative approaches and promises more efficient and flexible navigation. However, in highly dynamic environments employing different kinds of obstacle classes, safe navigation still presents a grand challenge. In this paper, we propose a semantic Deep-reinforcement-learning-based navigation approach that teaches object-specific safety rules by considering high-level obstacle information. In particular, the agent learns object-specific behavior by contemplating the specific danger zones to enhance safety around vulnerable object classes. We tested the approach against a benchmark obstacle avoidance approach and found an increase in safety. Furthermore, we demonstrate that the agent could learn to navigate more safely by keeping an individual safety distance dependent on the semantic information.