Test for self-driving motor vehicle

Abstract

A computer includes a processor and a memory, the memory storing instructions executable by the processor to collect control data indicating behavior of a vehicle controlled by a control program, collect virtual control data indicating behaviour of a virtual vehicle controlled by a virtual control program, and determine a difference between the control data and the virtual control data.

Claims

1. A system, comprising a computer including a processor and a memory, the memory storing instructions executable by the processor to: collect, from one or more vehicle sensors, control data indicating a trajectory of a vehicle controlled by a control program according to a specified traffic scenario; collect virtual control data indicating a trajectory of a virtual vehicle controlled by a virtual control program according to the specified traffic scenario by which the vehicle is controlled by the control program; determine a difference between the control data and the virtual control data; and based on the difference, adjust one or more parameters of the virtual control program.

2. The system of claim 1, wherein the vehicle includes a data importer and a vehicle response detector, wherein the data importer is designed to import sensor data from the vehicle, and the vehicle response detector is designed to analyze the sensor data to determine behavior of the vehicle, and wherein the virtual vehicle includes a virtual data importer and a virtual vehicle response detector, wherein the virtual data importer is designed to import virtual sensor data from the virtual motor vehicle, and the virtual vehicle response detector is designed to analyze the virtual sensor data to determine behavior of the virtual vehicle.

3. The system of claim 2, wherein the vehicle response detector is programmed to provide a dataset including a trajectory of the vehicle, and the virtual vehicle response detector is programmed to provide a dataset representing a virtual trajectory of the virtual vehicle, and wherein the instructions further include instructions to compare the dataset including the trajectory and the dataset including the virtual trajectory.

4. The system of claim 1, wherein the control software differs from the virtual control software.

5. The system of claim 1, wherein the control software is the same as the virtual control software, and the instructions further include instructions to determine a quantity representing the differences between the control data and the virtual control data.

6. The system of claim 1, wherein the instructions further include instructions to determine a quality criterion for the control software in a virtual driving environment.

7. The system of claim 1, wherein the instructions further include instructions to collect the control data in a physical driving environment.

8. The system of claim 1, wherein the virtual vehicle is a virtual model of the vehicle.

9. A method, comprising: collecting, from one or more vehicle sensors, control data indicating a trajectory of a vehicle controlled by a control program according to a specified traffic scenario; collecting virtual control data indicating a trajectory of a virtual vehicle controlled by a virtual control program according to the specified traffic scenario by which the vehicle is controlled by the control program; determining a difference between the control data and the virtual control data; and based on the difference, adjust one or more parameters of the virtual control program.

10. The method of claim 9, wherein the vehicle includes a data importer and a vehicle response detector, wherein the data importer is designed to import sensor data from the vehicle, and the vehicle response detector is designed to analyze the sensor data to determine behavior of the vehicle, and wherein the virtual vehicle includes a virtual data importer and a virtual vehicle response detector, wherein the virtual data importer is designed to import virtual sensor data from the virtual motor vehicle, and the virtual vehicle response detector is designed to analyze the virtual sensor data to determine behavior of the virtual vehicle.

11. The method of claim 10, wherein the vehicle response detector is programmed to provide a dataset including a trajectory of the vehicle, and the virtual vehicle response detector is programmed to provide a dataset representing a virtual trajectory of the virtual vehicle, and wherein the instructions further include instructions to compare the dataset including the trajectory and the dataset including the virtual trajectory.

12. The method of claim 9, wherein the control software differs from the virtual control software.

13. The method of claim 9, wherein the control software is the same as the virtual control software, and the method further includes determining a quantity representing the differences between the control data and the virtual control data.

14. The method of claim 9, further comprising determining a quality criterion for the control software in a virtual driving environment.

15. The method of claim 9, further comprising collecting the control data in a physical driving environment.

16. The method of claim 9, wherein the virtual vehicle is a virtual model of the vehicle.

Description

BRIEF DESCRIPTION OF THE DRAWING

(1) FIG. 1 is a schematic diagram of a system for testing a self-driving motor vehicle.

DETAILED DESCRIPTION

(2) In the present exemplary embodiment, the system 2 comprises a real motor vehicle 4, a real driving environment 6, a virtual motor vehicle 8, a virtual driving environment 10, a virtual environment generator 12, a communication module 14, real control software 16, virtual control software 18, a real data import system 20, a real vehicle response detection module 22, a virtual data import system 24, a virtual vehicle response detection module 26, and a central module 28.

(3) Said components can comprise hardware and/or software components for their tasks and functions described below.

(4) In the present exemplary embodiment, the real motor vehicle 4 is a car. In addition, in the present exemplary embodiment, the real motor vehicle 4 is embodied as a self-driving motor vehicle that can drive, steer and park without any input from a human driver. For this purpose, the real motor vehicle 4 comprises various sensors (not shown) for detecting the environment, and from the information obtained, can determine its position and the other road-users, work with the navigation software to drive to a destination, and avoid collisions on the way thereto.

(5) The real driving environment 6 is a cordoned-off test site, on which the real motor vehicle 4 can be tested.

(6) The virtual motor vehicle 8 is a software model of the real motor vehicle 4.

(7) The virtual driving environment 10 is, for instance, a simulation of the real driving environment 6, though which the virtual motor vehicle 8 can be navigated. In other words, the virtual driving environment 10 is a software model of the real driving environment 6.

(8) The virtual environment generator 12 imports real vehicle data such as, for example, sensor data, map and navigation data, or traffic data, and populates the virtual driving environment 10 with this data, for instance in order to simulate specific traffic situations.

(9) The communication module 14 is used for the data transfer between the real and virtual components. Thus both online and offline simulations can be performed.

(10) The real control software 16 controls the real motor vehicle 4, for instance when the real motor vehicle 4 is in the real driving environment 6.

(11) The virtual control software 18 may be identical to the real control software 16, or they may be different software versions.

(12) The real data import system 20 collects data from the real motor vehicle 4 during operation.

(13) The real vehicle response detection module 22 imports the data and analyzes it to determine, in the present exemplary embodiment, a real vehicle trajectory.

(14) During a simulation, for instance during a simulated journey of the virtual motor vehicle 8 through a virtual driving environment 10, the virtual data import system 24 collects data in a similar way to the real data import system 20.

(15) Similarly, the virtual vehicle response detection module 26 imports the data and analyzes it to determine, in the present exemplary embodiment, a virtual vehicle trajectory.

(16) The central module 28 imports the real vehicle trajectory and the virtual vehicle trajectory and compares one with the other.

(17) During operation, real tests are performed first, for example, in which the real motor vehicle 4 is moved through a real driving environment 6 under the open-loop/closed-loop control of the real control software 16.

(18) In this process, the real data import system 20 collects data, which the real vehicle response detection module 22 imports and analyzes to determine, for example, the real-world trajectory of the real motor vehicle 4.

(19) In a further step, virtual tests are performed, in which the virtual motor vehicle 8 is moved through a virtual driving environment 10, which is provided by the virtual environment generator, under the open-loop/closed-loop control of the virtual control software 18.

(20) In this process, the virtual data import system 24 collects data, which the virtual vehicle response detection module 26 imports and analyzes to determine, for example, the virtual world trajectory of the virtual motor vehicle 8.

(21) The central module 28 then compares the real-world trajectory with the virtual world trajectory and provides a comparison result.

(22) As an alternative to the present exemplary embodiment, the test can have a different sequence, for instance it is also possible to start with the virtual test followed by the real test, or the real and virtual tests are performed simultaneously, at least in part.

(23) The comparison result can be used to refine and improve a virtual test environment by automatically supplying the virtual test environment with real data. The comparison result can also be used for certification, homologation and/or autonomous driving approval. It can be employed for the back-to-back tests and/or regression tests of software updates in a continuous integration environment. In addition, the comparison result can be used to monitor the behavior of motor vehicles in the real world, and to detect deviations from normal behavior in good time.

(24) Hence the results from the virtual tests and the real tests can be automatically merged and analyzed.

LIST OF REFERENCES

(25) 2 system 4 real motor vehicle 6 real driving environment 8 virtual motor vehicle 10 virtual driving environment 12 virtual environment generator 14 communication module 16 real control software 18 virtual control software 20 real data import system 22 real vehicle response detection module 24 virtual data import system 26 virtual vehicle response detection module 28 central module