Autonomous vehicle path planning method based on deep convolutional neural network

Abstract

The invention relates to an autonomous vehicle path planning method based on a deep convolutional neural network. The method comprises the following steps: firstly, building a deep convolutional neural network model; secondly, acquiring a barrier grid map of a vehicle under different driving environments, and a corresponding reference path for building a sample database for the training, verification and testing of the deep convolutional neural network model; and finally, applying the trained and tested deep convolutional neural network model to the autonomous vehicle, inputting the barrier grid map and the reference path into the neural network in real time, generating a key sampling area, and sampling, planning and generating a path from the key sampling area by using a sampling-based path planning algorithm. Through adoption of the autonomous vehicle path planning method, the problem that the prediction error cannot be avoided in an end-to-end method is solved; the running track of an autonomous vehicle is more stable and controllable; the defect of lack in sensitivity and adaptivity of a method for selecting a path from a preset path set is overcome; the path planning flexibility is improved; and meanwhile the path quality is ensured.

Description

technical field [0001] The present invention relates to the field of deep convolutional neural network and automatic driving, and more specifically, relates to a method for path planning of automatic driving vehicle based on deep convolutional neural network. Background technique [0002] In recent years, unmanned driving technology has gradually become a research hotspot in universities and enterprises at home and abroad, and it has also attracted public attention because of its frequent appearance in commercial application tests. In the process of automatic driving of unmanned vehicles, the path planning subsystem of the autonomous driving system is the key basis for the automatic driving of unmanned vehicles, and it is the premise to ensure the safety and stability of unmanned vehicles and normal driving without collision. [0003] Due to its superior performance, deep convolutional neural networks have achieved great success in many fields such as image classification, o...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the defects in the above-mentioned prior art that the prediction error of the neural network affects the control of unmanned vehicles, the present invention provides a path planning method for automatic driving vehicles based on a deep convolutional neural network, using a deep convolutional neural network Automatically extract features from the reference path and obstacle grid map and get the key sampling area, and then use the sampling-based path planning algorithm to sample and generate paths from the key sampling area, thus overcoming the prediction error in the end-to-end method. Problems overcome to make the trajectory of autonomous vehicles more stable and controllable

Images