Adjusting the gain factor of an accelerometer for a motor vehicle

Abstract

Device for adjusting a gain factor (VF) of an acceleration controller (BR) for a motor vehicle, wherein: • the acceleration controller (BR) is configured to specify a target acceleration (SB) for the motor vehicle in a time step as a function of a target speed (SG) of the motor vehicle, an actual speed (IG) of the motor vehicle and the gain factor (VF), and • the device is configured to: • store the target speed (SG), the actual speed (IG) and the target acceleration (SB) specified therein as information for at least two time steps, • select a first subset (ET) of the information, • train a model (MU) as a function of the first subset (ET), wherein the model (MU) is configured• to predict an actual speed (IG) of a later time step from at least one stored actual speed (IG) and at least one stored target acceleration (SB), • to select a second subset (ZT) of the information, and • to adjust the gain factor (VF) depending on the second subset (ZT), the model (MU) and the acceleration controller (BR).

Description

[0001] The invention relates to a device and a method for adjusting a gain factor of an acceleration controller for a motor vehicle.

[0002] Within the context of this document, the term "automated driving" can refer to driving with automated longitudinal or lateral control, or autonomous driving with automated longitudinal and lateral control. The term "automated driving" encompasses automated driving at any level of automation. Examples of automation levels include assisted, partially automated, highly automated, and fully automated driving. These levels of automation were defined by the German Federal Highway Research Institute (BASt). In assisted driving, the driver continuously performs longitudinal or lateral control, while the system takes over the other function within certain limits. In partially automated driving (TAF), the system takes over longitudinal and lateral control for a certain period and / or in specific situations, whereby the driver must continuously monitor the system, as in assisted driving.In highly automated driving (HAD), the system takes over longitudinal and lateral control for a certain period without the driver needing to continuously monitor the system; however, the driver must be able to take over control of the vehicle within a certain timeframe. In fully automated driving (VAD), the system can automatically handle driving in all situations for a specific use case; no driver is required for this use case. The four automation levels mentioned above, as defined by the German Federal Highway Research Institute (BASt), correspond to SAE Levels 1 to 4 of the SAE J3016 standard (SAE - Society of Automotive Engineering). For example, highly automated driving (HAD) corresponds to Level 3 of the SAE J3016 standard according to BASt. Furthermore, SAE J3016 also includes SAE Level 5 as the highest level of automation, which is not included in the BASt definition.SAE Level 5 corresponds to driverless driving, where the system can automatically handle all situations like a human driver throughout the entire journey; a driver is generally no longer required.

[0003] All longitudinal control systems are based on an acceleration controller that calculates a target acceleration for the vehicle from a desired vehicle speed and the vehicle's actual speed. The longitudinal control of the vehicle is then regulated, at least in relation to this target acceleration.

[0004] Parameterizing the acceleration controller is very complex, as the acceleration controller must be adapted for each vehicle model, and factors such as the vehicle mass and various properties of the drivetrain have an influence on the parameterization.

[0005] The purpose of the invention is to simplify the parameterization of the acceleration controller.

[0006] The problem is solved by the features of the independent claims. Advantageous embodiments are described in the dependent claims. It should be noted that additional features of a dependent claim, without the features of the independent claim itself or only in combination with a subset of the features of the independent claim, can constitute a separate invention independent of the combination of all features of the independent claim, which can be the subject of an independent claim, a divisional application, or a subsequent application. This applies equally to technical teachings described in the description, which can constitute an invention independent of the features of the independent claims.

[0007] A first aspect of the invention relates to a device for adjusting a gain factor of an acceleration controller for a motor vehicle, in particular for an automated motor vehicle.

[0008] The acceleration controller is designed to specify a target acceleration for the vehicle in a time step, depending on a target speed of the vehicle, an actual speed of the vehicle and the amplification factor.

[0009] The longitudinal control of the vehicle then occurs at least as a function of the target acceleration. In particular, the target acceleration is specified as the desired acceleration by a drive or engine control unit. Alternatively or additionally, the target acceleration is processed further before being specified as the desired acceleration to the drive or engine control unit.

[0010] The device is designed to store the target speed, the actual speed, and the target acceleration specified depending on it as information for at least two time steps.

[0011] In particular, the device is designed to store the target speed, the actual speed and the target acceleration specified depending on it as tuples, so that the stored information also shows that the aforementioned data correspond to the same time step.

[0012] In particular, the device is designed to store the target speed, the actual speed, the target acceleration specified depending on it, and the respective time step as information for at least two time steps, so that a causal or temporal sequence of the aforementioned data can also be derived from the stored information.

[0013] Furthermore, the device is configured to select a first subset of the information, wherein the first subset comprises, in particular, at most 150 or 200 tuples of target speed, actual speed, and / or target acceleration. The invention is based on the understanding that the number of tuples is chosen such that processing under real-time conditions, i.e., with mandatory adherence to a deadline, is possible.

[0014] The device is configured to train a model depending on the first subset, wherein the model is configured to predict an actual speed of a later time step from at least one stored actual speed and at least one stored target acceleration.

[0015] The invention is based on the finding that the actual speed at the second time step can be predicted from the actual speed and the target acceleration at a first time step, taking into account the time difference between the first time step and a second time step following the first time step.

[0016] While the actual acceleration of a vehicle will often deviate from the target acceleration, as it depends not only on factors controllable by the vehicle (e.g., road gradient, signal propagation times within the vehicle, and / or system inertia), the actual vehicle speed is stored and thus known for several time steps. This allows the model to be trained retrospectively using a supervised learning process.

[0017] Furthermore, the device is configured to select a second subset of the information, wherein the second subset comprises, in particular, at most 20, 50, 100 or 150 tuples of target speed, actual speed and / or target acceleration.

[0018] Furthermore, the device is set up to adjust the gain factor depending on the second subset, the model and the acceleration controller.

[0019] The invention is based on the understanding that the choice of amplification factor has a strong influence on how quickly and with what quality the actual speed of the vehicle adjusts to a deviating target speed. For example, a very high amplification factor can ensure a rapid adjustment of the actual speed to the target speed, but a very high amplification factor combined with time delays poses a risk of oscillations.

[0020] The device is specifically designed to allow the model training and acceleration controller adjustment to be performed multiple times in order to iteratively converge at an optimal gain factor. For example, by appropriately selecting the frequency at which the model training and acceleration controller adjustment are performed, the optimum can be found with low computational power.

[0021] In an advantageous embodiment of the invention, the acceleration controller is configured to determine the target acceleration from the product of the amplification factor and the difference between the target speed and the actual speed.

[0022] In a further advantageous embodiment of the invention, the device is configured to store the information in a ring buffer, wherein the capacity of the ring buffer is limited to storing the information of a maximum of 5000 time steps.

[0023] A ring buffer continuously stores data over a certain period of time and overwrites it after a predetermined time has elapsed in order to free up storage space for new data.

[0024] In particular, the time difference between any two time steps is at most 20 ms, so that the ring buffer can store at most information from an interval of 100 s.

[0025] In a further advantageous embodiment of the invention, the device is set up to train the model by optimizing a first weighting factor and a second weighting factor in such a way as to minimize a prediction error of the model.

[0026] In particular, the optimization of the first and second weighting factors is performed using a Levenberg-Marquardt algorithm. The invention is based on the finding that the Levenberg-Marquardt algorithm converges very quickly for this problem compared to other optimization algorithms, which, in conjunction with other measures, enables the use of the invention in motor vehicles (i.e., "online," as opposed to "offline" training in a data center).

[0027] The first weighting factor determines the influence of at least one stored actual speed on the prediction. In particular, if the at least one stored actual speed includes more than just one actual speed, multiple first weighting factors can be used. For example, a separate first weighting factor can be used for each of the multiple actual speeds.

[0028] The second weighting factor determines the influence of at least one stored target acceleration on the prediction. In particular, if the at least one stored target acceleration includes more than just one, multiple second weighting factors can be used. For example, a separate second weighting factor can be used for each of the multiple target accelerations.

[0029] In a further advantageous embodiment of the invention, the device is configured to adjust the amplification factor by being configured to predict a state of the motor vehicle depending on the second subset, the model and the acceleration controller.

[0030] The state of the vehicle is, in particular, a description of the vehicle's actual dynamics and / or a description of control or target values ​​for vehicle systems that will influence the vehicle's dynamics in the future. For example, the state of the vehicle includes a target acceleration for the current time step, an actual speed for the current time step, and a target speed for the current time step. Additionally, the state of the vehicle may also include an actual speed and / or a target acceleration for at least one past time step.

[0031] In particular, since the complete state of the motor vehicle can only be described with great difficulty, the state of the motor vehicle in the present embodiment of the invention can only be described partially, for example by at least an actual speed of the motor vehicle, at least a target speed of the motor vehicle and / or at least a target acceleration of the motor vehicle.

[0032] Furthermore, the device is designed to adjust the gain factor in such a way as to minimize a control quality measure related to the condition of the motor vehicle.

[0033] The controller performance measure describes, in particular, a control deviation and / or a measure of passenger comfort.

[0034] In particular, the adjustment of the gain factor is carried out using a Levenberg-Marquardt algorithm. The invention is based on the finding that the Levenberg-Marquardt algorithm converges very quickly for this problem compared to other optimization algorithms, which, in conjunction with further measures, enables the use of the invention in motor vehicles.

[0035] The state of the motor vehicle includes, in particular, at least an actual speed of the motor vehicle and / or at least a target acceleration of the motor vehicle and / or at least a target speed of the motor vehicle in a time step.

[0036] For example, the model can be used to predict, starting from an initial state of the vehicle, how the target acceleration, target speed and actual speed of the vehicle will develop in future time steps if different values ​​are assumed for the gain factor of the acceleration controller.

[0037] In a further advantageous embodiment of the invention, the device is configured to store the information in a ring buffer, wherein the capacity of the ring buffer is limited to storing the information of a maximum of 5000 time steps, to train the model by optimizing a first weighting factor and a second weighting factor using a Levenberg-Marquardt algorithm such that a prediction error of the model is minimized, wherein the first weighting factor specifies an influence of the at least one stored actual speed on the prediction, and wherein the second weighting factor specifies an influence of the at least one stored target acceleration on the prediction, and to adjust the gain factor by predicting a state of the motor vehicle depending on the second subset, the model and the acceleration controller.and to optimize the gain factor using a Levenberg-Marquardt algorithm in such a way that a controller performance measure related to the state of the motor vehicle is minimized.

[0038] This advantageous embodiment combines all the features that make the invention so efficient that it allows the invention to be used directly in the motor vehicle despite the limited resources of automotive control units.

[0039] A second aspect of the invention relates to a method for adjusting a gain factor of an acceleration controller for a motor vehicle, wherein the acceleration controller is configured to specify a target acceleration for the motor vehicle in a time step depending on a target speed of the motor vehicle, an actual speed of the motor vehicle and the gain factor.

[0040] One step is to store the target speed, the actual speed, and the target acceleration specified depending on it as information for at least two time steps.

[0041] Another step in the process is selecting a first subset of the information.

[0042] A further step of the procedure is to train a model depending on the first subset, whereby the model is set up to predict an actual speed of a later time step from at least one stored actual speed and at least one stored target acceleration.

[0043] A further step in the process is selecting a second subset of the information.

[0044] Another step in the process is adjusting the gain factor depending on the second subset, the model, and the acceleration controller.

[0045] The foregoing descriptions of the device according to the first aspect of the invention also apply accordingly to the method according to the second aspect of the invention. Advantageous embodiments of the method according to the invention not explicitly described here or in the claims correspond to the advantageous embodiments of the device according to the invention described above or in the claims.

[0046] The invention is described below using an exemplary embodiment with the aid of the accompanying drawing.

[0047] This shows a device according to the invention for adjusting a gain factor VF of an acceleration controller BR for a motor vehicle.

[0048] The acceleration controller BR is designed to specify a target acceleration SB for the motor vehicle in a time step depending on a target speed SG of the motor vehicle, an actual speed IG of the motor vehicle and the gain factor VF.

[0049] Furthermore, the acceleration controller BR is set up to determine the target acceleration SB from the product of the gain factor VF and the difference between the target speed SG and the actual speed IG.

[0050] The device is designed to store the target speed SG, the actual speed IG and the target acceleration SB specified depending on it as information for at least two time steps.

[0051] In particular, the device is designed to store the information in a ring buffer RS, wherein the capacity of the ring buffer RS ​​is limited to storing the information for a maximum of 5000 time steps.

[0052] Furthermore, the device is configured to select a first subset ET of the information and to train a model MU depending on the first subset ET, wherein the model MU is configured to predict an actual speed IG of a later time step from at least one stored actual speed IG and at least one stored target acceleration SB.

[0053] In particular, the device is set up to train the model MU by optimizing a first weighting factor and a second weighting factor such that a prediction error of the model MU is minimized, wherein the first weighting factor specifies an influence of the at least one stored actual velocity IG on the prediction, and wherein the second weighting factor specifies an influence of the at least one stored target acceleration SB on the prediction.

[0054] Furthermore, the device is set up to select a second subset ZT of the information and to adjust the gain factor VF depending on the second subset ZT, the model MU and the acceleration controller BR, for example by using an optimization tool CU.

[0055] In particular, the device is configured to adjust the gain factor VF by configuring the device to predict a state of the motor vehicle depending on the second subset ZT, the model MU and the acceleration controller BR, and to adjust the gain factor VF in such a way that a controller performance measure related to the state of the motor vehicle is minimized.

[0056] The state of the motor vehicle includes at least an actual speed IG of the motor vehicle and / or at least a target acceleration SB of the motor vehicle in one time step.

Claims

[1] Device for adjusting a gain factor (VF) of an acceleration controller (BR) for a motor vehicle, wherein • the acceleration controller (BR) is set up to specify a target acceleration (SB) for the vehicle in a time step depending on a target speed (SG) of the vehicle, an actual speed (IG) of the vehicle and the gain factor (VF), and • the device is set up, • to store the target speed (SG), the actual speed (IG) and the target acceleration (SB) specified depending on it as information for at least two time steps, • to select a first subset (ET) of the information, • to train a model (MU) depending on the first subset (ET), wherein the model (MU) is set up to predict an actual speed (IG) of a later time step from at least one stored actual speed (IG) and at least one stored target acceleration (SB), • to select a second subset (ZT) of the information, and • to adjust the gain factor (VF) depending on the second subset (ZT), the model (MU) and the acceleration controller (BR). [2] Device according to claim 1, wherein the acceleration controller (BR) is configured to determine the target acceleration (SB) from the product of the gain factor (VF) and the difference between the target speed (SG) and the actual speed (IG). [3] Device according to one of the preceding claims, wherein the device is configured to store the information in a ring buffer (RS), wherein the capacity of the ring buffer (RS) is limited to storing the information for a maximum of 5000 time steps. [4] Device according to one of the preceding claims, wherein • the device is set up to train the model (MU) by optimizing a first weighting factor and a second weighting factor such that a prediction error of the model (MU) is minimized, • the first weighting factor specifies an influence of at least one stored actual speed (IG) on the prediction, and • the second weighting factor has an influence that specifies at least one stored target acceleration (SB) on the prediction. [5] Device according to any of the preceding claims, wherein the device is configured to adjust the gain factor (VF) by configuring the device to • to predict a state of the motor vehicle depending on the second subset (ZT), the model (MU) and the acceleration controller (BR), and • to adjust the gain factor (VF) in such a way as to minimize a control performance measure related to the condition of the motor vehicle. [6] Device according to claim 5, wherein the state of the motor vehicle comprises the actual speed (IG) of the motor vehicle and / or the target acceleration (SB) of the motor vehicle in at least one time step. [7] Device according to any of the preceding claims, wherein the device is set up, • to store the information in a ring buffer (RS), wherein the capacity of the ring buffer (RS) is limited to storing the information for a maximum of 5000 time steps, • to train the model (MU) by optimizing a first weighting factor and a second weighting factor using a Levenberg-Marquardt algorithm such that the prediction error of the model (MU) is minimized, where the first weighting factor specifies the influence of at least one stored actual velocity (IG) on the prediction, and where the second weighting factor specifies the influence of at least one stored target acceleration (SB) on the prediction, and • to adjust the gain factor (VF) by predicting a state of the motor vehicle depending on the second subset (ZT), the model (MU) and the acceleration controller (BR), and • to optimize the gain factor (VF) using a Levenberg-Marquardt algorithm in such a way that a controller performance measure related to the state of the motor vehicle is minimized. [8] Method for adjusting a gain factor (VF) of an acceleration controller (BR) for a motor vehicle, wherein • the acceleration controller (BR) is set up to specify a target acceleration (SB) for the vehicle in a time step depending on a target speed (SG) of the vehicle, an actual speed (IG) of the vehicle and the gain factor (VF), and • the procedure includes the following steps: • Storing the target speed (SG), the actual speed (IG) and the corresponding target acceleration (SB) as information for at least two time steps, • Selecting a first subset (ET) of the information, • Training a model (MU) depending on the first subset (ET), wherein the model (MU) is set up to predict an actual speed (IG) of a later time step from at least one stored actual speed (IG) and at least one stored target acceleration (SB), • Selecting a second subset (ZT) of the information, and • Adjusting the gain factor (VF) depending on the second subset (ZT), the model (MU) and the acceleration controller (BR).

Images