SUMMARY

This thesis focuses on a development of a vision system to aid an automated charging robot for electric vehicles (EVs) and plug-in hybrid vehicles (PHEVs). The automated charging robot needs to recharge EVs or PHEVs by inserting a standard electric charging plug into a car’s charging socket. With a use of computer vision and deep learning techniques, the vision system assists the robot by detecting and providing a charging inlet location on a vehicle. Additionally, the vision system identifies a condition of the charging port (open and close) with a variety of charging port colors and types. Object detection and depth estimation models are both used in the vision system using deep learning algorithms. The models are validated with standard evaluation metrics used in the object detection and the depth estimation. Based on experimental results and the studies, the vision system successfully identified a location between the robot and the charging inlet of the vehicle.