Limited Field of View-Driven Path Planning for Human Search in Indoor Environments

This study proposes an innovative and efficient approach to indoor robot navigation by integrating Limited Field of View Optimization with Room Segmentation. Our method effectively maximizes both sensor coverage and path planning efficiency, setting it apart from traditional topology-based navigation techniques that often fall short in complex indoor environments. Using ROS2 and Gazebo, we validate the algorithm in a simulated environment, where Voronoi Diagrams and Distance Transform Maps are employed for optimized node placement and path planning. Additionally, experiments conducted with a LiDAR- and HD camera-equipped robot in a real-world indoor setup demonstrate a notable 64% reduction in travel distance when compared to baseline models, highlighting the algorithm's strong potential to enhance human-robot interaction through more efficient coverage. Future work will explore further improvements in robustness across a variety of indoor settings and dynamic environments.