적외선 카메라를 이용한 자동차 페달의 딥러닝 인식 연구

초록

Recently, the harshness of testing performed by human has increased in indoor chassis dynamometer tests due to the application of time-consuming test methods such as single-charge driving distance tests for electric vehicles and real-road emission tests. To solve these problems, automotive test automation using pedal robots is in progress, but existing robots are expensive and take a lot of time to install. In this study, we developed a device that can automatically recognize the positions of the brake pedal and accelerator pedal, enabling quick installation and test start in a short time. Using this, a pedal robot placed at an arbitrary location can accurately recognize the pedal positions of the car and quickly start the test. In this paper, we conducted a deep learning study to recognize the positions of the two pedals using an infrared camera in a dark driver's seat environment.