Method for calculating driving strategy of pilotless automobile in urban traffic scene

Abstract

The invention relates to the field of machine learning, and discloses a method for calculating a driving strategy of a pilotless automobile in an urban traffic scene, which comprises the following steps: collecting data samples, and preprocessing the data samples to obtain a data set; constructing and training a multi-scale prediction convolutional neural network based on the data set; inputting an image acquired by the camera into a multi-scale prediction convolutional mental network, and acquiring a target category and confidence in a scene; identifying a traffic scene according to the target category; and calculating a driving strategy risk degree and moral intensity based on the traffic scene, and generating a driving strategy. According to the invention, the recognition precision and overall efficiency of the target object are improved, and the target category in the image is recognized while the precision of the recognized target is calculated.

Description

technical field [0001] The invention relates to the field of machine learning, in particular to a driving strategy calculation method for an unmanned vehicle in an urban traffic scene. Background technique [0002] With the rapid development of artificial intelligence technology and the continuous improvement of deep learning algorithms, autonomous driving technology is gradually moving towards the track of practical application. At present, some cities in China have allowed self-driving online car-hailing vehicles on the road. At the same time, in order to make self-driving cars better adapt to complex urban traffic scenarios and make ethical driving strategies in ethical dilemmas, it is necessary to More targeted detection of urban traffic scenes and traffic elements in the scenes. Traditional scene classification methods based on convolutional neural networks are all pursuing high-precision classification methods. They do not start from practical applications, but more s...

AI Technical Summary

Problems solved by technology

Traditional scene classification methods based on convolutional neural networks are all pursuing high-precision classification methods. They do not start from practical applications, but more specifically classify traffic scenes from traffic elements, and show the categories of finally recognized objects and their corresponding Confidence, and there is no research on the driving strategy of self-driving cars in ethical dilemmas

Images