COMPUTER-IMPLEMENTED METHOD FOR DETECTING A STEERING WHEEL INTERVENTION CONDITION, COMPUTER PROGRAM PRODUCT, DRIVING ASSISTANCE SYSTEM AND MOTOR VEHICLE

Description

[0001] This document describes a computer-implemented method for detecting a steering wheel intervention state, a computer program product, a driver assistance system and a motor vehicle.

[0002] Computer-implemented methods for detecting a steering wheel intervention state, computer program products, driver assistance systems and motor vehicles of the type mentioned above are known in the prior art.

[0003] Motor vehicles are increasingly being equipped with autonomous or semi-autonomous driving functions. Furthermore, motor vehicles often feature driver assistance systems that perform various specific tasks, such as adaptive cruise control, lane keeping assist, or lane tracking systems. With semi-autonomous driving functions or driver assistance systems, the driver must always be able to take control of the vehicle at any time. It is therefore important for the corresponding driver assistance systems to be able to accurately determine the status of any driver intervention in the vehicle's control devices, such as the brakes, accelerator pedal, or steering wheel, at any time.

[0004] In steering systems, for example, it is known to use capacitive sensors on the rim of a steering wheel to detect whether a driver is holding at least one hand on the steering wheel. It is also known to use a torque sensor to measure the forces exerted on the steering by a driver using the steering wheel. Both known systems have different disadvantages. For example, it is possible for a driver to touch the steering wheel but not exercise control over the vehicle. Furthermore, it is known that capacitive sensors have problems with drivers wearing gloves. Capacitive sensors are also relatively expensive. The fact that at least one hand is on the rim of a steering wheel therefore says nothing about whether the driver is currently controlling the vehicle.In addition, it is known that torque sensors that measure steering force can sometimes produce incorrect measurements or can be outsmarted, for example by hanging a weight on one side of the steering wheel.

[0005] DE 10 2009 028 647 A1 discloses a power steering system comprising a boost module that generates an estimated driver torque and a mixing module for determining a mixing value. The mixing value is based at least in part on a derivative of the estimated driver torque, and the mixing value is applied to a return torque of a steering wheel.

[0006] DE 10 2016 005 013 A1 shows a generic computer-implemented method and describes a steer-by-wire steering system for motor vehicles with a feedback actuator, wherein a hands-on / off detection is provided based on measured values ​​of the feedback actuator.

[0007] Furthermore, DE 10 2013 113 628 A1 describes a hands-on / off detection system for a vehicle steering wheel based on a frequency analysis of steering angle data or steering torque data.

[0008] The task therefore arises of improving computer-implemented methods for detecting a steering wheel intervention state, computer program products, driver assistance systems and motor vehicles of the type mentioned above in such a way that a more reliable detection of a driver taking over or relinquishing control is possible.

[0009] The object is achieved by a computer-implemented method for detecting a steering wheel intervention state according to claim 1, a computer program product according to the independent claim 9, a driver assistance system according to the independent claim 10 and a motor vehicle according to the independent claim 11. Further embodiments and developments are the subject of the dependent claims.

[0010] The following describes a computer-implemented method for determining a steering wheel intervention state of a driver of a motor vehicle, wherein the motor vehicle has at least one driving function intervening in a steering arrangement of the motor vehicle, wherein a steering torque at the steering arrangement is detected and a steering torque signal is generated therefrom, wherein in a first branch a steering torque shift (or a steering torque offset) is determined from the steering torque signal, wherein a Kalman filter is used, wherein the Kalman filter uses the steering torque signal and the steering torque shift as measured variables, wherein the Kalman filter filters the steering torque shift out of the steering torque signal in order to determine a steering torque estimate, wherein the steering torque estimate is compared with at least one limit value, wherein the comparison determines whether the driver is intervening in the steering arrangement.

[0011] The driving function can be a driver assistance function, e.g., a lane tracking or lane keeping function, or an automated or semi-automated function, e.g., a traffic jam assistant or a parking function. The method can also be used by the motor vehicle in multiple driving functions, including a fatigue detection system.

[0012] The steering torque can be detected or measured, for example, using a force sensor or a torque sensor arranged on the steering assembly. A corresponding force sensor or torque sensor can be functionally arranged between a steering column and a steering wheel of the motor vehicle's steering assembly and thus measure forces or torques applied to the steering column via the steering wheel. The steering torque signal can be a continuous signal or a signal sampled at a given sampling frequency.

[0013] Such sensors are inherently noisy due to external influences, meaning that determining the state of steering wheel intervention based solely on the steering torque signal can lead to incorrect conclusions. The noise can, for example, originate from the sensor itself (e.g., thermal noise) or from external influences, including vibrations. Furthermore, it is possible that a corresponding sensor has manufacturing and / or measurement tolerances and, among other things, emits asymmetric signals (right-left) or exhibits a static or quasi-static offset. Such influences can lead to measurement inaccuracies and thus to incorrect state determinations. For example, a certain measurement could lead to the conclusion that the driver is operating the steering wheel when this is actually not the case. If corresponding driver assistance functions are then deactivated, this could lead to dangerous traffic situations.

[0014] Any steering actuators that may be present and are used to implement corresponding driver assistance functions or automated driving functions are usually arranged on a steering gear and generally act on the steering gear, so that no torque is to be expected between the steering wheel and the steering column if only a corresponding steering actuator intervenes.

[0015] A Kalman filter is a mathematical method for the iterative estimation of parameters describing system states based on error-prone observations. The Kalman filter can be used to estimate system variables that cannot be directly measured while optimally reducing measurement errors. For dynamic variables, a mathematical model can be added to the filter as a constraint to account for dynamic relationships between the system variables. For example, equations of motion can help to accurately estimate changing positions and velocities together.

[0016] Because the Kalman filter uses both a steering torque signal and a steering torque offset, a clean signal in the form of a steering torque estimate can be generated, which can then be compared to at least one threshold value. It can be provided that there are certain torque value ranges that are considered plausible for steering wheel intervention, which do not fall below certain very low torques and do not exceed certain high torques that a driver cannot generate during normal driving.

[0017] The Kalman filter can have two states and two measured variables.

[0018] Furthermore, a distinction can be made between steering intervention and termination of a steering intervention with different limit values ​​or value ranges, so that in various embodiments one limit value, two limit values, three limit values ​​or four limit values ​​are used with which the steering torque estimate output by the Kalman filter is compared.

[0019] In a first further embodiment, it can be provided that the steering torque shift is determined by means of a band-pass filter.

[0020] Using a band-pass filter, it is possible to isolate certain signal frequency ranges that are typical for shifts or offsets. Such offsets usually vary slowly, which is why their dynamic range is usually low-frequency.

[0021] In a further refinement, it can be provided that the Kalman filter is an adaptive Kalman filter with a covariance matrix that is adaptively adjusted to compensate for both a static steering torque shift and a dynamic steering torque shift.

[0022] A possible way to perform adaptation in adaptive extended Kalman filters (AEKF) can be found in "S. Akhlaghi," "N. Zhou," and "Z. Huang," "Adaptive adjustment of noise covariance in Kalman filters for dynamic state estimation." in IEEE Power and Energy Society General Meeting (PESGM 2017), 16-20 July 2017, Chicago, IL, USA .

[0023] An adaptive Kalman filter has the advantage that its process model can be adapted to existing conditions, since, especially in the case of steering systems, environmental conditions cannot be expected to be constant or nearly constant. Automotive steering systems are subject to numerous disturbances, such as rapidly changing external forces. Such influences can include changes in reaction forces due to changes in the road surface, road irregularities, crosswinds, road gradients, and the like, which are subject to constant change during a journey.

[0024] In a further further embodiment, it can be provided that a time derivative of the steering torque signal is formed in a second branch in order to determine a steering torque variation, wherein the steering torque variation is compared with at least one limit value, wherein the comparison determines whether the driver intervenes in the steering arrangement.

[0025] By providing a second branch, which in one embodiment can be processed in parallel to the first branch, in which a time derivative of the steering torque signal is formed, a sudden increase or decrease in a steering torque can be used as a signal that a driver is engaging or releasing a steering wheel.

[0026] In a further refinement, it can be provided that a low-pass filter is applied to the steering torque signal and / or the time derivative of the steering torque signal before and / or after the time derivative of the steering torque signal.

[0027] By using low-pass filters before and / or after the time derivative, it is possible to smooth the time derivative of the steering torque signal and eliminate high-frequency influences that can be caused, among other things, by vibrations. Such vibrations can be caused by the road, engine, and / or tires, among other things.

[0028] In a further refinement, it can be provided that a determination of a driver intervention in the steering arrangement is made from the first branch or from the second branch.

[0029] The two branches make it possible to receive and process signals from the steering torque itself as well as from the change in the steering torque.

[0030] In a further refinement, it can be provided that the first branch generates a first output signal, wherein the second branch generates a second output signal, wherein the first output signal and the second output signal are combined, wherein it is determined from the combined output signals whether the driver intervenes in the steering arrangement.

[0031] By combining the two output signals, it is possible to make a robust prediction about the driver's intervention in the steering arrangement.

[0032] In a further refinement, it can be provided that the combined output signals are debounced by means of a debouncing function, wherein it is determined from the combined debouncing output signals whether the driver intervenes in the steering arrangement.

[0033] Using a debouncing function, it is possible to eliminate any existing time differences in the evaluations of the two branches as well as other causes that can lead to bouncing of the resulting signal.

[0034] A first independent subject matter relates to a device for determining a steering wheel intervention state of a driver of a motor vehicle, wherein the motor vehicle has at least one driving function intervening in a steering arrangement of the motor vehicle, wherein a steering torque sensor is arranged for detecting a steering torque on the steering arrangement and for generating a steering torque signal therefrom, wherein a controller is provided which is configured to determine a steering torque shift from the steering torque signal in a first branch, wherein a Kalman filter is provided, wherein the Kalman filter uses the steering torque signal and the steering torque shift as measured variables, wherein the Kalman filter is configured to filter the steering torque shift from the steering torque signal and to determine a steering torque estimate, wherein the controller is configured tocomparing the steering torque estimate with at least one limit value, the comparison determining whether the driver intervenes in the steering arrangement.

[0035] In a first further embodiment, a band-pass filter is provided for determining the steering torque shift.

[0036] In a further refinement, it can be provided that the Kalman filter is an adaptive Kalman filter with a covariance matrix that is adaptively adjustable to compensate for both a static steering torque shift and a dynamic steering torque shift.

[0037] In a further refinement, it can be provided that the controller is configured to form a time derivative of the steering torque signal in a second branch in order to determine a steering torque variation, wherein the controller is configured to compare the steering torque estimate with at least one limit value, wherein the comparison determines whether the driver intervenes in the steering arrangement.

[0038] In a further further embodiment, it can be provided that the control has at least one low-pass filter for application to the steering torque signal and / or the time derivative of the steering torque signal.

[0039] In a further further embodiment, it can be provided that the control is configured to determine an intervention of the driver in the steering arrangement from the first branch or from the second branch.

[0040] In a further further embodiment, it can be provided that the controller is configured to generate a first output signal from the first branch and a second output signal from the second branch, wherein the controller is configured to combine the first output signal and the second output signal, wherein the controller is further configured to determine from the combined output signals whether the driver intervenes in the steering arrangement.

[0041] In a further refinement, it can be provided that the controller is configured to debounce the combined output signals by means of a debouncing function, wherein the controller is configured to determine from the combined debouncing output signals whether the driver intervenes in the steering arrangement.

[0042] A further independent subject matter relates to a computer program product comprising a permanent computer-readable storage medium on which instructions are embedded which, when executed by at least one computing unit, cause the at least one computing unit to be configured to carry out the method of the aforementioned type.

[0043] The method can be carried out in a distributed manner on one or more computing units, so that certain method steps are carried out on one computing unit and other method steps are carried out on at least one further computing unit, whereby calculated data can be transmitted between the computing units if necessary.

[0044] Another independent subject matter relates to a driver assistance system of a motor vehicle, comprising an actuator (torque generator), a steering torque sensor and a controller, wherein the controller comprises a processor and a computer program product of the type described above.

[0045] A further independent subject matter relates to a motor vehicle having a computer program product of the type described above and / or a driver assistance system of the type described above.

[0046] Further features and details will become apparent from the following description, in which at least one embodiment is described in detail—possibly with reference to the drawings. Identical, similar, and / or functionally identical parts are provided with the same reference numerals. The following schematically show: Fig. 1 shows a plan view of a motor vehicle with a driver assistance system, and Fig. 2 shows a flow chart of the method.

[0047] Fig. 1 shows a plan view of a motor vehicle 2 with a driver assistance system 4 (framed in dashed lines).

[0048] Motor vehicle 2 is located in lane 6, which is marked with lane markings 8.

[0049] The driver assistance system 4 of the motor vehicle 2 is a lane tracking system that serves to keep the motor vehicle 2 in lane 6. The driver assistance system 4 is equipped with a camera 10, a steering actuator 12, and a steering torque sensor 14.

[0050] The controller 16 has a processor 18 and a non-volatile memory 20, wherein in the memory 20 represents a computer program product which, when loaded and executed by the processor 18, carries out the method described below (see Fig. 2 ) carried out.

[0051] The driver assistance system 4 uses data from the camera 10 to predict necessary interventions in the components of the motor vehicle 2. In this case, the focus is primarily on interventions in a steering assembly 22, which, in addition to the steering actuator 12, includes, among other things, a steering column 24 and a steering wheel 26. The steering torque sensor 14 measures a torque at the steering column 24.

[0052] The steering actuator 12 acts on a steering gear 27 into which the steering column 24 opens and via which steering forces are distributed to wheels of a front axle of the motor vehicle 2.

[0053] The driver assistance system 4 must always determine a steering intervention state of the steering arrangement 22. This includes, on the one hand, determining whether the driving function can operate without driver intervention and, on the other hand, whether the driver is currently operating the steering wheel 26 or not. The first analysis includes an analysis of the traffic situation and the driving state of the motor vehicle 2. Based on this, the driver assistance system 4 must decide whether the driver must assume control of the motor vehicle 2 or not. Relevant traffic situations can include, for example, visibility, road width, road type, road curvature, construction sites, and dangerous situations, for example, caused by nearby motor vehicles. In addition, the driver assistance system 4 must be able to recognize a driver's intention, for example, an intention to change lanes, a turning or evasive maneuver, or a position correction of the motor vehicle 2 in the lane.

[0054] Fig. 2 shows a flowchart of the procedure.

[0055] A steering torque T is output from the steering torque sensor 14 and processed by the controller 16. In a first branch 28, a steering torque shift T off is determined using a bandpass filter 30. The steering torque shift T off and the steering torque T are fed to an adaptive Kalman filter 32.

[0056] The adaptive Kalman filter 32 has a process model and a measurement model that makes several assumptions: 1. Process model: The assumptions for the process model of the adaptive Kalman filter 32 are:

[0057] A steering torque T and a steering torque shift T off applied by the driver are almost constant between two consecutive iterations k, k+1 of the procedure and changes in the steering torque T and the steering torque shift T off are represented by uncorrelated white noise w T and w off: T k + 1 = T k + w T T off , k + 1 = T off , k + w T , off 2. Measurement model: The assumptions for the measurement model of the adaptive Kalman filter 32 are:

[0058] An estimated measure of steering torque T̂ k+1 results from a steering torque T minus the steering torque shift T off plus measurement noise v T , an estimated measure of the steering torque T̂ off,k +1 results from the steering torque shift T off plus measurement noise v off . T ^ k + 1 = T k + 1 − T off , k + 1 + v T T ^ off , k + 1 = T off , k + 1 + v off ,

[0059] The filtering process of the adaptive Kalman filter 32 can be divided into two steps: prediction and estimation / correction. During the prediction step, an a priori estimate for the system states T̂ , T̂ off are calculated. During the estimation / correction step, these are compared with the measured values ​​T, T off and corrected accordingly.

[0060] A correction of the steering torque shift should only be carried out during the estimation / correction step if the amount of the measured steering torque shift (i.e. the output signal from the band-pass filter) is smaller than a certain upper limit, which can be, for example, between 0.1 and 0.3 Nm.

[0061] The adaptive Kalman filter 32 has a covariance matrix for describing the noise terms, which is adaptively adjusted to dynamic shifts, e.g. caused by changes in road conditions, engine vibrations and the like.

[0062] One way to perform the adaptation of the covariance matrix Q k For adaptive extended Kalman filters (AEKF), see "S. Akhlaghi," "N. Zhou," and "Z. Huang," "Adaptive adjustment of noise covariance in Kalman filters for dynamic state estimation," in IEEE Power and Energy Society General Meeting (PESGM 2017), 16-20 July 2017, Chicago, IL, USA : Q k + 1 = α Q k + 1 − α K k + 1 d k + 1 d T k + 1 K T k + 1 with: Q k : covariance matrix α : Forgetting factor (between 0 and 1) d k : remaining error (the difference between the actual measurement and the estimated measurement) K k : Kalman gain factor

[0063] In this case, an adaptive extended Kalman filter is used instead of an adaptive Kalman filter. The adaptation can be accelerated by applying the following rules for adapting the covariance matrix: Q k used: Q k + 1 = α Q k + 1 − α K k + 1 d k + 1 d T k + 1 K T k + 1 + K k + 1 L k + 1 L T k + 1 K T k + 1 L k + 1 = K I T s d k + 1 + d k with: K I : Integrating factor T s : Sample time

[0064] The result of the filtering process of the adaptive Kalman filter 32 is a filtered steering torque T KF , which is then compared with threshold values ​​in a comparison instance 34, from which a statement about driver intervention in the steering wheel 26 can be determined.

[0065] In a second branch 36, the steering torque T is filtered using a low-pass filter 38, then derived and then filtered again using a low-pass filter 42 in order to remove high-frequency disturbances in the steering torque T as well as in the time derivative of the steering torque Ṫ to eliminate F.

[0066] The filtered time derivative Ṫ F of the steering torque is fed to a comparison instance 44, which determines from this whether there is a steering intervention by the driver or not.

[0067] The comparison instances 34 of the first branch 28 and 44 of the second branch 36 each generate an output signal, the first branch 28 an output signal T 28 and the second branch 36 an output signal T 36 . The output signals T 28 , T 36 are fed to a debouncing function 46, which consolidates the state of the steering intervention by a driver with the aid of a comparison instance 48 and outputs a steering intervention state parameter PL.

[0068] Although the subject matter has been illustrated and explained in detail by means of exemplary embodiments, the invention is not limited by the disclosed examples, and other variations may be derived therefrom by those skilled in the art. It is therefore clear that numerous variations exist. It is also clear that the exemplary embodiments mentioned are merely examples and should not be construed as limiting the scope, possible applications, or configuration of the invention in any way.Rather, the preceding description and the description of the figures enable the person skilled in the art to implement the exemplary embodiments in concrete terms, whereby the person skilled in the art, with knowledge of the disclosed inventive concept, can make various changes, for example with regard to the function or arrangement of individual elements mentioned in an exemplary embodiment, without departing from the scope of protection defined by the claims and their legal equivalents, such as a further explanation in the description. List of reference symbols

[0069] 2Motor vehicle 4Driver assistance system 6Lane 8Lane marking 10Camera 12Steering actuator 14Steering torque sensor 16Controller 18Processor 20Non-volatile memory 22Steering assembly 24Steering column 26Steering wheel 27Steering gear 28First branch 30Band-pass filter 32Adaptive Kalman filter 34Comparison instance 36Second branch 38Low-pass filter 40Derivative function 42Low-pass filter 44Comparison instance 46Debouncing function 48Comparison instance TSteering torque TF Low-pass filtered steering torque T KF Filtered steering torque T off Steering torque shift T 28 , T 36 Output signal Ṫ F filtered time derivative of the steering torque PL steering intervention state

Claims

1. Computer-implemented method for determining a steering wheel engagement state of a driver of a motor vehicle (2), wherein the motor vehicle (2) at least one in a steering arrangement (22) of the motor vehicle (2) engaging driving function, wherein a steering torque at the steering arrangement (22) detected and a steering torque signal (t) is generated therefrom, wherein in a first branch (28) a steering torque shift (toff) from the steering torque signal (t) is determined, wherein a kalman - filter (32) is used, wherein the kalman - filter (32) as measured variables the steering torque signal (t) as well as the steering torque shift (toff), wherein the kalman - filter (32) the steering torque shift (t) from the steering torque signal (t) uses filtered to determine a steering torque estimate (TKF), wherein the steering torque estimate (TKF) with at least one limit value is compared, wherein by means of the comparison it is determined whether the driver in the steering arrangement (22) engages.

2. Computer implemented method of claim 1, wherein the steering torque shift is determined by a band pass filter (30).

3. Computer implemented method of claim 1 or 2, wherein the Kalman filter is an adaptive Kalman filter (32) having a covariance matrix adaptively adapted to compensate for both a static steering torque shift and a dynamic steering torque shift.

4. Computer implemented method according to any one of the preceding claims, wherein in a second branch (36) a time derivative of the steering torque signal (t) is formed, in order to determine a steering torque variation (t ˙f), wherein the steering torque variation (t ˙f) with at least one limit value is compared, wherein by means of the comparison it is determined whether the driver in the steering arrangement (22) engages.

5. Computer implemented method according to claim 4, wherein a low-pass filter (38, 42) is applied to the steering torque signal (T) and / or the time derivative (T ˙F) of the steering torque signal (T) before and / or after the time derivative of the steering torque signal (T).

6. Computer implemented method of claim 5, wherein a determination of driver intervention in the steering assembly (22) is made from the first branch (28) or the second branch (36).

7. Computer implemented method of any of the preceding claims 4 to 6, wherein the first branch (28) generates a first output signal (T28), the second branch (36) generates a second output signal (T36), wherein the first output signal (T28) and the second output signal (T36) are combined, the combined output signals (T28, T36) being used to determine whether the driver is intervening in the steering assembly (22).

8. Computer implemented method of claim 7, wherein the merged output signals (T28, T36) are debounced by means of a debouncing function (46), wherein the merged debounced output signals (T28, T36) are used to determine whether the driver is intervening in the steering assembly (22).

9. Computer programme product comprising a computer readable non-volatile storage medium (20) having embedded thereon instructions which, when executed by at least one computing unit (18), cause the at least one computing unit (18) to be adapted to perform the method of any preceding claim.

10. Driving assistance system of a motor vehicle, comprising an actuator (12), a steering torque sensor (14) and a controller (16), wherein the controller (16) comprises a processor (18) and a computer programme product according to claim 9.

11. Motor vehicle with a computer programme product according to claim 9 and / or a driving assistance system (4) according to claim 10.