The Implementation of an Autonomous Navigation System on a Self-Driving CART

Various methodologies used to achieve autonomous navigation have led to multiple outcomes focused mainly on increasing navigation efficiency, reducing cost, and improving safety. This project revolves around developing an autonomous navigation system for a vehicle to navigate autonomously. The system comprises multiple modules that work cohesively to achieve autonomous navigation, such as the vision sensors, mapping, localization, and path planning modules. The vision sensors module had an A2M7 RP LiDar and a ZED2i stereo camera. The jetson nano microcomputer was chosen to achieve less cost and computational power as it has proven efficient performance in AI deep learning operations. Hector SLAM was selected as the core algorithm to accomplish the task. In addition, Robot Operation Software (ROS) was used as the system's middleware. The mapping module used the LiDar and ZED2I camera data to generate a map of the environment. The system was tested, and the results show an accurate path with minimal diversions and collisions. It is vital to mention that other teams constructed the vehicle's hardware and propulsion design, as this was a multi-disciplinary project. However, our research focuses on how the autonomous navigation system was achieved. The proposed system has the potential to be adopted in various robot applications such as exploration, transportation, and surveillance.