Field test method and device for self-driving vehicles

Abstract

The field test method and device of the self-driving vehicle provided by the present invention collect and obtain the real traffic scene data in advance, and obtain the current field environment data when the self-driving vehicle to be tested is tested on the field, and obtain the corresponding traffic scene data from the real traffic scene data set. The traffic scene data, using the traffic scene data obtained by the matching as the test scene data, automatically controls the movement trajectory of the traffic participants, sends the target task to the self-driving vehicle to be tested, and then realizes the test of the self-driving vehicle to be tested. Using real traffic scene data as test data for real vehicle testing is more in line with the real traffic environment and can improve the authenticity and validity of the test results. Moreover, the entire field test process does not require manual control of each traffic participant, which improves the accuracy of the test process. safety. Based on the analysis of a large amount of road data, the probability of occurrence of different types of scenarios can be quantified, thereby improving the accuracy of test results.

Description

Technical field [0001] The present invention relates to an automatic driving technology, and more particularly, a method and apparatus for field testing involves autonomous vehicle. Background technique [0002] Autonomous vehicles, also known as unmanned vehicles, computers or driving a vehicle wheeled mobile robot, through a computer system to achieve intelligent unmanned vehicle. Autonomous vehicles in the real roads, need to deal with complex and changing traffic scene. Currently, for field testing method for automatic driving of the vehicle are the following: 1 is manually set using the test data, a function test on an autonomous vehicle in a controlled venue; as used for the artificial test data set, can not be sufficiently. reflect the complexity of real traffic environment, resulting in less accurate test results. 2. After the traffic accident scene reconstruction based on data for testing autonomous vehicles, due to the accident data reflect only extremely dangerous scen...

AI Technical Summary

Problems solved by technology

At present, there are the following several methods for field testing of self-driving vehicles: 1. In order to utilize the test data set manually, the function test of self-driving vehicles is carried out in a controlled field; Reflect the complexity of the real traffic environment, resulting in poor accuracy of test results

2. After reconstructing the scene based on the traffic accident data, the self-driving vehicle is tested, because the accident data only reflects the extremely dangerous scene, the test results are not complete

The process of building simulation test scenarios and vehicle simulation models is very complicated, and affected by specific software, some test scenarios are difficult to realize

At the same time, there are application problems such as the difficulty of evaluating the test accuracy in virtual testing

Images