Convolutional Neural Network Behavioral Cloning in Self-Driving

EasyChair Preprint 948

Abstract

To reduce the cost of the intelligent driving technology of intelligent car in a specific environment. An unmanned system with behavior cloning method is designed, which based on convolutional neural network. This method is a process that mimics human learning, unmanned driving by predicting the state of the vehicle through convolutional neural networks, ordinary visual sensors are used to obtain environmental information. First, behavioral cloning method based on convolutional neural network is studied. Then, an intelligent vehicle based on Linux system Raspberry Pi is designed, two sets of experiments are designed to verify the feasibility of the method. Finally, the desired model is applied to the first “DIY Robocars KuaiKai” Race Categories. The results show that the behavioral cloning method can realize the self-driving technology of intelligent car in a specific environment.

Keyphrases: Convolutional Neural Network, Raspberry Pi, behavioral cloning, intelligent vehicle, self-driving, vision sensor

@booklet{EasyChair:948,
  author    = {Haolong Fu and Jin Zhao and Axing Xi},
  title     = {Convolutional Neural Network Behavioral Cloning in  Self-Driving},
  howpublished = {EasyChair Preprint 948},
  year      = {EasyChair, 2019}}