Enhancing Automatic Electric Vehicle Charging: A Deep Learning Approach with YOLO and Feature Extraction Techniques

Phasuwut Chunnapiya Porawat Visutsak ^*

• King Mongkut's University of Technology North Bangkok, Bang Sue District, Thailand

This research addresses the challenge of automating electric vehicle (EV) charging in Thailand, where five distinct EV charging plug types are prevalent. We propose a deep learning approach using YOLO (You Only Look Once) to accurately identify these plug types, enabling robots to perform charging tasks efficiently. The study evaluates four YOLO versions (V5s, V6s, V7, and V8s) to determine the optimal model for this application. Our results demonstrate that YOLO V8s achieves the highest accuracy with a Mean Average Precision (mAP) of 0.95, while YOLO V7 exhibits superior performance in certain real-world scenarios. This research contributes to the development of automated EV charging systems by providing a robust and accurate model for detecting all five types of EV charging plugs used in Thailand. The model's ability to accurately detect and classify EV charging plugs paves the way for the design of automated charging robots, addressing a key challenge in EV charging infrastructure and promoting the wider adoption of electric vehicles.