Design and Integration of a Multi-Sensor System for Enhanced Indoor Autonomous Navigation

This study explores the forefront of indoor autonomous robotics, combining cutting-edge computational technologies with precise sensor systems. The NVIDIA Jetson Xavier NX serves as the powerful computational core, efficiently processing complex data. This central unit is enhanced by two critical sensors: the Intel RealSense D435 for depth perception, and a LiDAR unit for detailed environmental mapping, an Inertial Measurement Unit (IMU) to ensure accurate path adherence, all integrated seamlessly through the Robot Operating System (ROS). A key aspect of our research is the implementation of the YOLOv5 model for recognizing indoor traffic signals, utilizing the Jetson NX's prowess in PyTorch with CUDA. This feature is vital for the robot's navigation, enabling it to interpret and react to visual cues in complex indoor settings. Our approach significantly enhances the robot's navigational capabilities by utilizing a combination of advanced route planning and traffic light detection. The system employs the A* pathfinding algorithm to efficiently traverse between predefined coordinates. At each designated node, it employs the YOLOv5 model to detect traffic lights. The robot continues seamlessly upon detecting a green light or if there is no traffic light, whereas it stops 40cm before the light. Our goal is to showcase the effectiveness of combining multiple sensory technologies in autonomous robots. This system, when combined with the robot's depth sensing, advanced object detection via the YOLOv5 model, and LiDAR mapping, significantly streamlines the navigation process. Preliminary results indicate enhanced navigational precision and quicker response times, representing a step forward in sophisticated indoor autonomous navigation. This approach emphasizes the value of integrating different technologies, including efficient navigation systems, to achieve advanced and reliable autonomous movement, marking a significant contribution to the field of robotics technology.