Collision Avoidance in Mini Autonomous Electric Vehicles Using Artificial Potential Fields for Outdoor Environment

The rapid advancement of technology is driving the transition toward Society 5.0, where intelligent transportation systems enhance safety, efficiency, and sustainability. One of the biggest challenges in transportation is the high frequency of vehicle accidents, with approximately 80% attributed to driver error. To mitigate this, Advanced Driver Assistance Systems (ADAS) have been developed to improve vehicle autonomy and reduce accidents. This research proposes a potential field-based collision avoidance system for autonomous vehicle navigation, where the vehicle and obstacles act as positive poles, repelling each other, while the target destination serves as a negative pole, attracting the vehicle. Experimental results demonstrate a GPS positioning error of 1.55 m with a 66% success rate and LiDAR sensor accuracy of 96.4%, exceeding the required 95% threshold. Obstacle avoidance was tested with two safety thresholds (2 m and 2.5 m) across single- and two-obstacle scenarios. The 2 m threshold resulted in shorter travel distances (16.406 m vs. 16.535 m for 2.5 m) and faster completion times (19.036 s vs. 19.144 s), while the 2.5 m threshold provided greater clearance. GPS accuracy was significantly influenced by HDOP values and satellite count, with lower HDOP improving trajectory precision. The system successfully adjusted its trajectory in response to obstacles, ensuring effective real-time navigation.