Research on Automatic Driving System Based on the Integration of Vision and Satellite Convolutional Neural Network

Abstract

Due to high dynamic range, high overlap, low concentricity and limited computation of the driverless intelligent vehicles, it is hard to implement global path planning and local adjustment under multi-objects and multi-obstacle state. To deal with the above problems, this paper presents a fresh idea that is to transfer the main object of intelligent drive from the car to the road and by using dynamic virtual scees and edge computation to construct the smart road which applies the smart car to the road. Based on this, this paper proposes a path planning method that integrates the vision and satellite convolutional neural network for automatic drive system and explores the distortion of visual image and the redundant processing mechanism of automatic drive. What's more, this paper structures the multi-dimensional data compensatory strategies in view of convolutional neural network and boosts the optimal identification of signs in intelligent roads for the intelligent car. Ultimately, this paper is to achieve intelligent global path planning by fusing satellite navigation and vision navigation.