Self-Driving Car Using Machine Learning and Neural Network

Engineering notes

Key topics: self-driving car, self-driving, autonomous vehicle, control. See the paper for implementation details and experimental results.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Abstract: Self-driving cars are autonomous vehicles that can run on their own without human intervention and could represent a technological revolution in the next decade. This work presents the development of a low-cost prototype of a miniature selfdriving car model using simple and out-of-the-box technology. The idea is to use only camera modules to create self-driving cars that move between destinations with minimal human intervention. The purpose of this project is to accelerate the process of driving a car through automation. The results of this project will definitely reduce the number of car accidents that occur today. This project uses NVIDIA Jetson Nano as the main controller and has a processing power of about 500 gigaflops. The controller is mounted on the car platform itself and performs all the calculations needed to drive the car on board. The car uses a trained convolutional neural network (CNN) that predicts all the parameters that the car needs to drive smoothly. These are directly connected to the main steering mechanism and the output of the deep learning model determines the steering angle of the vehicle. Keywords: CNN, NVIDIA Jetson Nano, Self-driving, automation, deep learning

Links and sources

• Official / arXiv page
• PDF from original source

Comments