Method for testing scene recognition function of automobile ADAS

Abstract

The invention discloses a method for testing an ADAS scene recognition function of an automobile, and the method comprises the steps: enabling a vehicle to carry out ADAS function testing in a specified line and different scenes, and obtaining vehicle CAN data, a driving video and a fault scene when the ADAS function fails; after all tests are completed, the number of faults of each ADAS function in different scenes and the number of times of ADAS function faults caused by each scene are counted, the fault probability of each ADAS function in different scenes is obtained, and the minimum number of times of tests needed when each ADAS function breaks down in different scenes is obtained; adjusting the type of a test scene, a specified line and the test times of each ADAS function in different scenes; and the ADAS function test is carried out again. According to the method, the types and the number of the test routes and the test scenes are optimized according to the test result, so that the test route is shortened, the test time is shortened, and the fault detection rate of the ADAS function is improved.

Description

technical field [0001] The invention relates to the technical field of automobile testing, in particular to a method for testing automobile ADAS scene recognition functions. Background technique [0002] With the widespread application of automotive ADAS intelligent driving assistance systems, there are more and more controllers on the vehicle, software algorithms and codes are becoming more and more complex, and the requirements for functional safety are also increasing accordingly. Therefore, the scope of automotive reliability needs to be improved. Expand accordingly. From the perspective of the automotive aftermarket, the number of recalls and the number of vehicles related to automotive intelligent systems have been on the rise in recent years, and new patterns of automotive defects have emerged, including scene recognition defects, software defects, and information security defects. With regard to scene recognition defects, due to insufficient scene testing and verifi...

AI Technical Summary

Problems solved by technology

As far as road tests are concerned, most car companies usually combine the assessment of ADAS functionality and environmental (scenario) adaptability with reliability road tests during the development process. More than 1 million kilometers, the cycle is usually 3-6 months, the rectification cycle is long after the problem is discovered, and there is a risk of insufficient verification before listing

On the other hand, the number of test prototype vehicles is large and the cost is high.

Most of the currently public ADAS road test methods are equivalent to blind tests, lack of scene and route design rules, and the purpose of the test is not clear enough, resulting in insufficient detection rate and detection speed of development problems

Images