Granted Patent B2

US 12,254,415 · App. 17/559,828 · Granted Mar 18, 2025

Inventors: Sebastian Bannenberg (Paderborn, DE); Fabian Lorenz (Paderborn, DE); Rainer Rasche (Paderborn, DE)

Assignee: dSPACE GMBH

G06N3/088B60W50/06B60W60/00G05B13/027G06F11/3664G06F11/3688G06F11/3692G06N3/045G06N3/08

Abstract

Claims (42)

1. A computer-implemented method for approximating a subset of test results of a virtual test of a device, the method comprising:

providing a data set defining a state space, wherein each state is formed by a parameter set of driving situation parameters for which state one or more actions are feasible to attain another parameter set from the parameter set, wherein each parameter set has at least one environment parameter describing an environment of a motor vehicle and at least one EGO parameter describing a state of the motor vehicle; and

performing an approximation step in which a function value of at least one further parameter set is approximated using an artificial neural network,

wherein the at least one further parameter set lies within a line-shaped edge area of a target range,

wherein if the function value of the approximated at least one further parameter set is greater than or equal to a predetermined threshold value, the at least one further parameter set is identified as associated with the subset of the test results, the subset representing a boundary between critical and non-critical test results,

wherein if the function value of the at least one further parameter set is less than the predetermined threshold value, the artificial neural network carries out at least one further approximation step starting from the respectively last approximated further parameter set until the function value of another parameter set is greater than or equal to the predetermined threshold, and

wherein the subset of the test results is used, at least partially, for an autonomous guidance of the motor vehicle.

2. The computer-implemented method according to claim 1 ,

wherein, from the plurality of parameter sets of the driving situation parameter, a startup parameter set is selected,

wherein in the approximation step, the function value of each adjacent parameter set attainable from the startup parameter sets by an action is approximated using the artificial neural network,

wherein a selection step is performed in which the parameter set approximated in the approximation step with a lowest or highest function value is selected, and

wherein if the function value of the selected parameter set is less than the predetermined threshold value, at least one further selection step is performed based on the respectively last selected parameter set until the function value of a selected parameter set is greater than or equal to the predetermined threshold value.

3. The computer-implemented method according to claim 1 , wherein the plurality of parameter sets of driving situation parameters is generated by the artificial neural network or by a simulation.

4. The computer-implemented method according to claim 1 ,

wherein the artificial neural network has four hidden layers each with 128 neurons and an ELU activation function, and

wherein a factor γ of 0.8 is used for the attenuation of the function value of the further approximation steps.

5. The computer-implemented method according to claim 1 ,

wherein a startup parameter set is selected from the plurality of parameter sets of the driving situation parameters,

wherein if the function value of the further parameter set is less than the predetermined threshold value, another artificial neural network rewards the further parameter set approximated by the artificial neural network and the artificial neural network is adapted on the basis of the reward, and

wherein the artificial neural network adapted in this manner performs at least one further approximation step starting from the last approximated parameter set until the function value of another parameter set is greater than or equal to the predetermined threshold value.

6. The computer-implemented method according to claim 5 ,

wherein the artificial neural network has four hidden layers each having 256 neurons and a PRELU activation function,

wherein the further artificial neural network has four hidden layers with 256 neurons each and one ELU activation function, and

wherein the artificial neural network and the other artificial neural network apply an Adam optimization method.

7. The computer-implemented method according to claim 1 , wherein the EGO parameters comprise a speed of the motor vehicle and the environmental parameters comprise a speed of another motor vehicle and a distance between the motor vehicle and the further motor vehicle.

8. The computer-implemented method according to claim 7 , wherein the plurality of driving situation parameters, in particular the speed of the motor vehicle and the speed of the further motor vehicle, are generated within a predetermined definition range by a random algorithm.

9. The computer-implemented method according to claim 1 , wherein the function underlying the function value is a safety target function having a numerical value which at a safety distance of ≥V FELLOW ×0.55 between the motor vehicle and the further motor vehicle has a minimum value, in a collision between the motor vehicle and the other motor vehicle has a maximum value, and at a safety distance of ≤V FELLOW ×0.55 between the motor vehicle and the other motor vehicle has a numerical value, which is greater than the minimum value.

10. The computer-implemented method according to claim 9 ,

wherein for approximating the numerical value range of each function underlying the function value, a separate artificial neural network is used,

wherein individual hyperparameters of each artificial neural network are stored in a database.

11. The computer-implemented method according to claim 1 , wherein the function underlying the function value is a comfort target function or energy consumption target function which has a numerical value, which has a minimum value in the event of no change in the acceleration of the motor vehicle, has a maximum value in a collision between the motor vehicle and the further motor vehicle, and in a change in the acceleration of the motor vehicle, has a numerical value between the minimum value and the maximum value as a function of the amount of change in the acceleration.

12. A non-transitory computer-readable data carrier with a program code of a computer program for carrying out the method according to claim 1 , when the computer program is executed on a computer.

13. A test unit for approximating a subset of test results of a virtual test of a device, the test unit comprising:

a component configured to provide a data set defining a state space, wherein each state is formed by a parameter set of driving situation parameters, for which state one or several actions are carried out in order to attain a further parameter set from the parameter set, wherein each parameter set has at least one environmental parameter describing the environment of a motor vehicle and at least one EGO parameter describing the state of the motor vehicle; and

an artificial neural network which performs an approximation step in which a function value of at least one further parameter set is approximated,

wherein the at least one further parameter set lies within a line-shaped edge area of a target range,

wherein if the function value of the approximated at least one further parameter set is greater than or equal to a predetermined threshold value, the at least one further parameter set is identifiable as associated with a subset of test results, the subset representing a boundary between critical and non-critical test results,

wherein if the function value of the at least one other parameter set is less than the predetermined threshold value, the artificial neural network is configured to perform at least one further approximation step starting from the respectively last approximated further parameter set until the function value of another parameter set is greater than or equal to the predetermined threshold value, and

wherein the subset of the test results is used, at least partially, for an autonomous guidance of the motor vehicle.

14. The test unit according to claim 13 ,

wherein the device is embodied by a control unit, and

wherein a driving situation underlying the approximation of the test results of the virtual test of the control unit is a lane change of another motor vehicle to a lane of the motor vehicle using the plurality of driving situation parameters.

Assignments (2)

Priority Claims (3)

Continuity (2)

References Cited (34)