Collision Avoidance for Autonomous Wheelchair Using Integrated SLAM and YOLO

Navigation and obstacle avoidance are fundamental components of autonomous systems in wheelchairs. These capabilities require the perception of the surrounding environment to determine safe paths. Simultaneous Localization and Mapping (SLAM) is an important technology in the development of these perception systems, that enables the creation of maps based on environmental data around the wheelchair. Light Detection and Ranging (LiDAR) is a widely used sensor for indoor mapping, but has limitations in detecting objects at certain height levels and non-uniform objects. On the other hand, RGB-D cameras can detect both static and dynamic objects beyond the limits of LiDAR, but have a narrower range and blind spots in mapping. This paper presents a collision avoidance system using the RPLiDAR A1M8 for environmental mapping with the RTAB-Map approach and the Intel RealSense D435i for obstacle detection based on the YOLO (You Only Look Once) model. Therefore, the integration of data from both sensors improves the performance of navigation and collision avoidance in wheelchairs.