Semantic-Enhanced Path Planning for Safety-Centric Indoor Robots Navigation

摘要

This paper presents an innovative approach for indoor mobile robot navigation, focusing on environments like Ambient Assisted Living spaces (AAL), smart homes, and factories, where humans and robots coexist. The need for effective space and traffic management in these areas is critical for ensuring safety and mobility. Our novel Semantic Path Planning algorithm improves upon traditional grid mapping by using a multi-dimensional array in the map file, where varying intensities indicate occupancy levels. The planner imposes penalties on free cells near specific static objects based on user preferences, facilitating autonomous robot movement in designated areas without creating rigid barriers. A standout feature of our algorithm is its ability to prioritize specific zones frequented by vulnerable groups like the elderly, or areas designated for rest, pets, or children's play. By enhancing the A* path planner with semantic and geometric data, our approach enables the management of these zones, leading to a comprehensive grid map for optimal path-finding. Moreover, this methodology is promising for multi-robot systems with differing navigation abilities and access rights. It also improves robot localization by focusing on unique environmental landmarks, thereby enhancing tracking accuracy and aiding in swift localization recovery. This approach is ideal for safe, efficient, and context-aware navigation in dynamic shared human-robot environments.