Abstract:The self-driving trolley created in this thesis uses cameras and ultrasonic sensors to obtain roadway information, and a deep learning based target recognition algorithm to find out which are the targets in the data obtained, so that the trolley can drive itself on a simulated roadway with functions such as obstacle avoidance and traffic signal recognition. Originally the car used a Raspberry Pi 3b+, but here the jetson nano, which is better than the Raspberry Pi 3b+, is used to implement it.

A RASPBERRY PI SELF-DRIVING CART BASED ON OPENCV AND DEEP LEARNING .