Dynamic scene information recording and simulation playback method, device, apparatus and medium

Abstract

The present invention relates to a dynamic scene information recording and simulation playback method and a dynamic scene information recording and simulation playback device. The method includes thefollowing steps that: sensor data are collected, local coordinate-to-global coordinate transformation is performed on the sensor data, so that absolute position information is generated; a model and ascene are established according to the absolute position information, so that the restoration of a road environment is realized, global coordinate-to-local coordinate transformation is performed on the sensor data, so that relative position information can be generated, the driving process of a vehicle is simulated based on the relative position information, so that the restoration of the drivingscene of a user can be realized; and the vehicle is planned and controlled according to planning control information and the relative location information, so that an improved planning control algorithm can be verified. The device comprises a preprocessing module, a simulation module and a planning control module. With the dynamic scene information recording and simulation playback method and device of the invention adopted, a fixed regression test scene can be formed; the scene is not only stable, but also can reproduce a current fault; and the reproduction of the fault and the restoration of the scene are not affected even through the planning control algorithm changes.

Description

technical field [0001] The present application relates to the field of simulation and verification of intelligent driving vehicles, and more specifically, relates to a method and device for recording and simulation playback. Background technique [0002] In order to enhance the stability of intelligent driving vehicles, a large number of real vehicle road tests are required. At present, when a problem is found in the road test, the tester will hand over the road test data to the R&D staff for processing. After the R&D personnel analyze the problem by analyzing the data recorded in the road test and find the root cause of the problem, the tester needs to go back to the scene where the problem occurred to test to see if the problem at that time has been solved. [0003] There are following problems in prior art: [0004] 1. The prior art only collects and records the position of the object relative to the sensor. Therefore, once the control algorithm changes, the data of th...

AI Technical Summary

Problems solved by technology

Therefore, when developing an improved planning control algorithm to avoid accidents or improve vehicle performance and want to verify the improved planning control algorithm, it can only be achieved through road measurements, wasting a lot of manpower and material resources

[0005] 2. Due to frequent changes in the actual road environment

Therefore, due to the influence of many uncertain factors in the real environment, it is difficult to reproduce the exact same scene as when the vehicle has a problem

Even in the road test, using the improved control algorithm and the fault problem of the vehicle no longer occurs, it cannot be guaranteed that the fault problem is really solved

[0006] 3. After the control algorithm is updated multiple times, the new control algorithm may cause some functions of the old control algorithm to fail or be affected

Repeatedly building the road measurement environment is not only time-consuming, labor-intensive and error-prone

Moreover, the environment for road testing is very unstable, which can easily lead to distortion of test results, making it difficult to find new defects in the planning control algorithm under the new version and difficult to truly avoid old defects, resulting in serious car accidents after the vehicle is on the road

[0007] Therefore, existing simulation and verification methods are flawed

Images