Method and device for estimating a steering torque in a steering system of a vehicle

The method employs a machine learning algorithm to estimate steering torque from electric motor data, addressing the limitations of torsion bar sensors by reducing costs and complexity in vehicle steering systems while maintaining accurate torque measurement.

WO2024240487A1PCT designated stage expired Publication Date: 2024-11-28ROBERT BOSCH GMBH
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
PCT/EP2024/062547
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-24
Filing Date
2024-05-07
Publication Date
2024-11-28

AI Technical Summary

Technical Problem

Steering systems in vehicles rely on torsion bar sensors for torque measurement, which require calibration and maintenance and may not always provide accurate signals, prompting the need for an alternative method to estimate steering torque effectively.

Method used

A computer-implemented method using a machine learning algorithm that processes input signals from an electric motor's movement variables to estimate the steering torque applied by an operator, potentially dispensing with the need for a torsion bar sensor, thereby reducing costs and installation space.

Benefits of technology

This approach allows for accurate estimation of steering torque without a torsion bar sensor, reducing costs and complexity, while also enabling the use of a single-channel sensor instead of a two-channel system, enhancing measurement accuracy and plausibility checks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure EP2024062547_28112024_PF_FP_ABST
    Figure EP2024062547_28112024_PF_FP_ABST
Patent Text Reader

Abstract

The invention relates to a computer-implemented method and a device (1) for estimating a steering torque in a steering system of a vehicle and / or machine.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] Description

[0002] title

[0003] Method and apparatus for estimating a steering torque in a of a vehicle

[0004] The invention relates to a method for estimating a steering torque in a steering system of a vehicle. Furthermore, the invention relates to a device for estimating a steering torque in a steering system of a vehicle.

[0005] State of the art

[0006] Vehicle steering systems are a critical component of modern vehicles, enabling the driver to steer the vehicle safely. Steering wheels and / or pedal controls are traditionally used to initiate steering movements and control the vehicle. Furthermore, joystick control is an alternative method to traditional steering wheel and pedal control. There are various types of joysticks that can be used to control vehicles, including one-hand and two-hand joysticks.

[0007] There are different types of steering systems, including mechanical, hydraulic, electric and electro-hydraulic steering systems.

[0008] Examples of mechanical steering systems include rack and pinion steering systems, recirculating ball steering systems, and / or worm and pinion steering systems. Hydraulic steering systems use hydraulic pressure to steer the wheels. Electric steering systems, on the other hand, use an electric motor to assist and / or perform the steering movements. Electro-hydraulic steering systems combine the advantages of both systems by combining a hydraulic pump with an electric motor.

[0009] In modern vehicles, the so-called steer-by-wire (SbW) steering system is often preferred. This is a vehicle control system in which a mechanical steering system is replaced by electronic and / or electromechanical systems. In this system, a steering wheel or other steering control element is not directly connected to the wheels via a steering spindle. Instead, the steering is controlled via sensors and / or steering actuators that convert the driver's steering movement into electronic signals and send them to a control unit via at least one cable or wirelessly. The control unit then controls the wheels via electric motors and / or other drives. Steer-by-wire systems enable greater precision and adaptability in the steering and can help reduce the space required in the vehicle.

[0010] Modern steering systems often use a torsion bar sensor to measure the steering angle initiated by the driver and / or operator via the steering control. The torsion bar sensor is a mechanical component used to measure rotational movements and / or torques and / or forces. The torsion bar sensor usually consists of a metal rod or bar that is deformed by a rotational movement. The deformation of the bar can be measured electronically and / or mechanically and provides information about the rotational movement and / or force acting on the bar.

[0011] A typical application for torsion bar sensors is force measurements in motor vehicles. There, they can be used, for example, to measure wheel loads, monitor axle loads, and / or measure steering forces. Torsion bar sensors are also used in other areas such as medical technology, robotics, and industrial engineering. Since the torsion bar sensor is an additional component that requires calibration and / or maintenance, and may not always provide a correct measurement signal, there are increasing efforts to provide information about rotational movement and / or torque and / or force in alternative or complementary ways.

[0012] The invention is therefore based on the object of specifying an improved and in particular hardware-optimized method and an improved and in particular hardware-optimized device.

[0013] The object is achieved by a method for estimating a steering torque in a steering system of a vehicle according to the features of patent claim 1. Furthermore, the object is achieved by a device for estimating a steering torque in a steering system of a vehicle according to the features of patent claim 10.

[0014] Disclosure of the invention

[0015] In a first aspect, a computer-implemented method for estimating a steering torque in a steering system of a vehicle and / or a machine is provided. The method comprises at least the following steps: providing a machine learning algorithm; providing at least one input signal comprising information about at least one movement variable of an electric motor of an operating element of the steering system; and estimating the steering torque applied by an operator via the operating element using the machine learning algorithm based on the provided input signal.

[0016] It is understood that the steps according to the invention, as well as further optional steps, do not necessarily have to be performed in the order shown, but can also be performed in a different order. Furthermore, further intermediate steps can be provided. The individual steps can also comprise one or more sub-steps without thereby departing from the scope of the method according to the invention. In a second aspect, a device for estimating a steering torque in a steering system of a vehicle and / or a machine is specified.The device comprises a provision device configured to provide a machine learning algorithm and at least one input signal comprising information about at least one movement variable of an electric motor of an operating element of the steering system; and an evaluation and calculation device configured to estimate and / or approximate the steering torque applied by an operator via the operating element by applying the machine learning algorithm based on the provided input signal.

[0017] One advantage of the invention is that by estimating the steering torque, the use of a torsion bar sensor in the respective steering system can be fundamentally dispensed with. According to the invention, it is possible to estimate the steering torque without necessarily having to measure the steering torque using a torsion bar sensor. This can reduce the costs of such a steering system. Furthermore, the torsion bar sensor is eliminated as a component, thus freeing up more installation space. The method and device according to the invention also have a positive effect on the production costs of a steering system.

[0018] In principle, the torsion bar sensor can also be retained in the steering system in question and supplemented by the inventive estimation of the steering torque. This can be preferred, for example, to verify the plausibility of measurement results from the torsion bar sensor and / or to increase the measurement accuracy of the torsion bar sensor. Furthermore, the additional and / or supplementary estimation of the steering torque allows a single-channel torsion bar sensor to be installed instead of a two-channel torsion bar sensor, so that the costs for such a torsion bar sensor and also the complexity of an evaluation device for the torsion bar sensor can be reduced. The electric motor is preferably a steering wheel actuator (SWA). In other words, the electric motor is preferably a control element actuator.The at least one input signal preferably comprises information about at least one movement variable of the electric motor of the control element, which in particular indicates how the electric motor moves and / or behaves and / or the current (actual) position of a rotor of the electric motor relative to a stator of the electric motor. The at least one input signal preferably comprises a plurality of input data items about the electric motor. The input data preferably comprises movement data of the electric motor. The electric motor can in principle also comprise a transmission, although this is preferably not a steering transmission of the steering system.

[0019] In the sense of the invention, a vehicle is understood to be a cargo vehicle and / or a passenger transport vehicle and / or a transport vehicle and / or a motor vehicle and / or a watercraft and / or an aircraft and / or a military vehicle and / or a crane vehicle and / or an excavator vehicle and / or a tractor and / or a tracked vehicle and / or a construction vehicle and / or a mobile hydraulic vehicle. Furthermore, the method and / or the device according to the invention can be used for estimating a steering torque in a steering system of a machine, in particular an industrial machine, and / or a tool. In other words, the method and / or the device according to the invention can also be used in other applications, in particular non-vehicle applications, in which, for example, steering inputs are made to control a tool head or the like.For example, the method and / or device according to the invention can be advantageous in the manual or semi-automatic control and / or steering of machines by an operator and / or by an automated function.

[0020] In particular, the method and / or device according to the invention is used in a steer-by-wire (SbW) steering system. In principle, the method and / or device according to the invention can also be used in other steering systems.

[0021] In one embodiment, the machine learning algorithm comprises an Explainable Boosting Machine (EBM) algorithm and / or a Generalized Additive Model (GAM) algorithm and / or a regression-based algorithm. In principle, other machine learning algorithms, such as neural networks and / or Gradient Boosted Tree algorithms and / or k-Nearest Neighbor algorithms and / or decision tree algorithms, can also be used.

[0022] A regression-based algorithm is easy to understand and particularly suitable for processing input signals and / or input data that exhibit a linear relationship to one another. In particular, such a regression-based algorithm is designed to determine a relationship between the input signals and / or input data based on a linear regression.

[0023] An Explainable Boosting Machine (EBM) algorithm is preferably a machine learning model designed to make predictions based on input data and / or input signals. EBM algorithms belong to the family of boosting algorithms, which create a stronger model by combining multiple weak learning models. Unlike many other machine learning models, an EBM algorithm offers greater interpretability because it is based on a decision rule structure that allows the prediction results to be presented in a human-readable form. This means it is easier to understand how and why the model arrived at a particular prediction. Another advantage of EBM algorithms is that they are robust to outliers and missing data and scale well when using categorical and / or numeric variables.Therefore, it is often used in applications that require high interpretability and accuracy.

[0024] A Generalized Additive Model (GAM) algorithm is a statistical model used to model a nonlinear relationship between a dependent variable and one or more independent variables. Unlike linear models, which can only model a linear relationship between variables, GAMs allow the modeling of complex nonlinear relationships. GAMs are based on the assumption that a nonlinear relationship between variables can be modeled using a combination of smooth functions. Smoothing functions can be, for example, splines and / or local regressions. The estimation of the smooth functions is performed using optimization algorithms that find the best combination of smooth functions. One advantage of GAMs is that they are highly interpretable, as they can represent the relationship between the dependent and independent variables in a graphical form.This can be very helpful in interpreting the results.

[0025] In one embodiment, the GAM algorithm determines a fit function for a plurality of data points based on a plurality of cumulative spline functions, in particular using at least one regularization term. These can be b-spline functions, for example. The respective spline function preferably describes a respective shape function for a respective data point and / or for a respective input signal. The regularization term can preferably be used to approximate a data curve and / or signal curve of the at least one input signal and / or the input data. A preferred GAM algorithm can also determine non-linear data relationships between the input signals and / or input data. Such a GAM algorithm is powerful and yet has a high degree of traceability compared to other machine learning algorithms.

[0026] Such a GAM algorithm can be given by the following formula: g(E[y]) = ßo + fi(xi) + f2(x2) + ... + fm(x m ) {Equation 1} where g(E(Y) is the fitted input signal function, ßo is a data offset to the at least one input signal, and f m (x m) is an m-th spline function for an m-th input signal. In one embodiment, the EBM algorithm is based on a GAM algorithm and is or is extended by a data tree structure-based approach. The EBM algorithm is therefore a data tree-based approach in which the input signals and / or the input data are bootstrapped and the at least one input signal or input data is aggregated. The EBM algorithm preferably comprises gradient boosting. The EBM algorithm is preferably trained by round-robin training. This is preferably based on series of restricted data trees. Furthermore, during training, residuals are preferably calculated from a respective input signal mean value. The EBM algorithm is scaling via the learning rate, which is considered to be very low.

[0027] Round-robin training is a machine learning method in which multiple models are trained simultaneously by applying them sequentially to the same dataset. In round-robin training, multiple models are trained by applying them sequentially to the same training dataset in a rotating cycle. This means that each model is trained once in turn before the next model is trained. This way, all models are trained evenly and prevents a model from always seeing the same training dataset and thus overfitting. After each training session, the model's weights are saved, and the next model starts training with those weights. This ensures that each model has a similar starting position and doesn't randomly start with different weights.Round-robin training can be particularly useful when it's difficult to find the best training approach for a given problem or when training data is limited. Using multiple models can reduce the risk of overfitting and improve overall model performance.

[0028] In one embodiment, the machine learning algorithm is trained based on training data about the at least one input signal or movement data of the electric motor of the control element of the steering system. The training data preferably comprise information about a large number of movement data items of the electric motor. The movement data can, for example, be determined at least partially by sensors, for example via a position and / or attitude sensor of the electric motor, which determines a position of the rotor relative to the stator. Alternatively or additionally, the movement data can also be provided by temporal differentiation of a sensor-detected measurement signal. The training of the machine learning algorithm is preferably carried out separately and / or in parallel for each piece of movement information about the movement of the electric motor.

[0029] In one embodiment, the at least one input signal comprises rotor position information and / or rotor speed information and / or rotor acceleration information and / or rotor jerk information and / or motor target torque information and / or information about a gradient of the motor target torque and / or actual motor torque information and / or information about a gradient of the actual motor torque. Particularly preferably, the respective information is recorded as a time series, so that a plurality of respective data points are available over a predetermined time, which the machine learning algorithm can evaluate. The machine learning algorithm is also trained, preferably feature-specifically, based on the data points.

[0030] In one embodiment, the at least one input signal is filtered, in particular by applying a moving average and / or a low-pass filter. In principle, other signal filters and / or data filters are also conceivable, preferably to exclude outliers and / or erroneous data points from the data set from the outset.

[0031] In one embodiment, the control element comprises a steering wheel and / or a joystick and / or a steering lever and / or a steering switch. The control element can vary depending on the type of vehicle and / or machine and / or steering technology. By way of example, the control element can comprise: a steering wheel, which is arguably the most commonly used steering control element in vehicles, typically comprising a frame with handles that the driver grasps and turns to steer the vehicle. It can preferably be (a) a steering wheel of a steer-by-wire steering system; (b) a joystick, which is an alternative to the steering wheel used in some vehicles and / or steering systems. The joystick resembles an aircraft control wheel and can preferably enable intuitive control; (c) a steering lever, as a steering control element, which is primarily used in commercial vehicles and / or agricultural machinery.The steering lever may be mounted either laterally or centrally in the driver's cab; and / or (d) a steering switch by which a vehicle and / or a machine and / or a steering system can be operated via at least one switch, for example on a console and / or on a dashboard.

[0032] The invention further relates to a control unit of a steering system and / or a vehicle, comprising at least one processor and / or an evaluation and control device configured to at least partially execute the method according to the invention in any desired embodiment. The control unit and / or the evaluation and control device need not execute all steps of the method, but can also execute only some of the steps of the method. The control unit can be designed as a smart control unit (SCU) of a steering system, also referred to as a steering SCU. The control unit can be designed as a control unit of a steering wheel actuator (SWA) and / or a control element actuator. The control unit can be a control unit of an electromechanical power steering system, also referred to as an EPS (Electric Power Steering) system.Alternatively, the control unit may be included in or formed by a central control and / or a central computer of a motor vehicle and / or a machine.

[0033] Furthermore, the invention relates to a steering system of a vehicle and / or a machine, comprising an operating element for applying a steering torque by an operator, an electric motor coupled to the operating element in a movement-transmitting and / or torque-transmitting manner; and a device according to the invention and / or a control unit according to the invention.

[0034] The invention also claims a computer program with program code for executing at least parts of the method according to the invention in one of its embodiments when the computer program is executed on a computer. In other words, the invention provides a computer program (product) comprising instructions that, when executed by a computer, cause the computer to execute the method / steps of the method according to the invention in one of its embodiments.

[0035] According to the invention, a computer-readable data carrier with program code of a computer program is also proposed for executing at least parts of the method according to the invention in one of its embodiments when the computer program is executed on a computer. In other words, the invention relates to a computer-readable storage medium comprising instructions which, when executed by a computer, cause the computer to execute the method / steps of the method according to the invention in one of its embodiments.

[0036] The described designs and further training courses can be combined as desired.

[0037] Further possible embodiments, further developments and implementations of the invention also include combinations of features of the invention described previously or below with regard to the embodiments that are not explicitly mentioned.

[0038] Short description of the drawings

[0039] The accompanying drawings are intended to provide a further understanding of embodiments of the invention. They illustrate embodiments and, in conjunction with the description, serve to explain principles and concepts of the invention. Other embodiments and many of the noted advantages will become apparent upon review of the drawings. The illustrated elements of the drawings are not necessarily drawn to scale relative to one another.

[0040] They show:

[0041] Fig. 1 is a schematic flow diagram of an inventive

[0042] procedure;

[0043] Fig. 2 is a schematic flow diagram of an exemplary

[0044] Machine learning algorithm;

[0045] Fig. 3A is a schematic diagram comparing a measured steering torque and a steering torque estimated according to the invention; and

[0046] Fig. 3B is a schematic diagram of a time course of a torque error of the steering torque estimated according to the invention.

[0047] In the figures of the drawings, the same reference symbols designate the same or functionally equivalent elements, parts or components, unless otherwise stated.

[0048] Figure 1 shows a schematic block diagram of a computer-implemented method and device 1 for estimating a steering torque in a steering system of a vehicle and / or a machine.

[0049] In any embodiment, the method can be carried out at least partially by the device 1, which for this purpose can comprise several components not shown in detail, for example, one or more provision devices and / or at least one evaluation and computing device. It is understood that the provision device can be designed jointly with the evaluation and computing device or can be different from it. Furthermore, the system can comprise a storage device and / or an output device and / or a display device and / or an input device.

[0050] According to the invention, the computer-implemented method comprises at least the following steps:

[0051] In a step S1, a machine learning algorithm 100 is provided. The machine learning algorithm comprises, for example, an Explainable Boosting Machine (EBM) algorithm, as shown schematically in a highly simplified and incomplete manner in Fig. 2. Alternatively or additionally, the machine learning algorithm comprises a Generalized Additive Model (GAM) algorithm and / or a regression-based algorithm.

[0052] In a step S2, at least one input signal is provided, which comprises information about at least one movement variable of the electric motor of an operating element of the steering system. The at least one input signal comprises rotor position information feati and / or rotor speed information featj and / or rotor acceleration information feats and / or rotor jerk information featn -4 and / or a motor target torque information feat n -3 and / or information about a gradient of the engine target torque feat n -2 and / or an actual engine torque information feat n -i and / or information about a gradient of the actual engine torque feat n . The features feat n -4 to feat n -i are not shown in Fig. 2, but are merely indicated by three dots (...).

[0053] In a step S3, the steering torque applied by an operator via the control element is estimated by the particularly trained machine learning algorithm on the basis of the provided at least one input signal.

[0054] In Fig. 2, a training procedure of the machine learning algorithm

[0055] 100, in this case an EBM algorithm is shown schematically in simplified form. The training is preferably carried out with n-features, feati to feat n , is used as a basis. In a first iteration step, a small, preferably non-complex data tree is trained based on the first feature feati. This data tree can preferably only use feature feati. By boosting, a residual value of the data tree is preferably updated. The process then preferably continues in an equivalent manner with feature featj. In the first iteration step, a small, preferably non-complex data tree is trained based on the second feature featj. This data tree can preferably only use feature featj. By boosting, a residual value of this data tree is preferably updated. The process preferably continues in the same way with the features feats to feat nThis procedure creates a round-robin path through all features feati to feat n Each of the generated data trees can preferably only consider a single feature. The learning rate of the method is so small that it preferably does not matter in which order the individual features are located feati to feat. n Rather, a large number of such iteration steps are executed, for example, up to 10,000 iteration steps or more. New data trees are preferably generated for each feature and each iteration step.

[0056] After completing the iteration steps i, for each feature feati to feat na number i of data trees are generated, each trained only on the basis of the respective feature. If these data trees are then summed feature by feature, a diagram 200, 202 is obtained for each feature. The respective diagram 200, 202 is generated by checking, based on the respective data tree, what the respective data tree estimates for each input value of the diagram for each output value.

[0057] Ultimately, one obtains preferably n diagram for the characteristics feati to feat n, so that the underlying data trees from the training are no longer required for further processing. Preferably, each diagram or each underlying data tree was trained in parallel. The resulting model or the resulting, trained machine learning algorithm 100 thus comprises the n diagrams for each of the n features. Based on a Generalized Additive Model (GAM) algorithm, a feature function for each feature can now be replaced and / or supplemented by a diagram generated as previously. Since these are modified functions, they are denoted by the index p instead of the index m in Equation 1. The index p preferably corresponds to the previously used index n. g(E[y]) = ßo + f(xi) + f2(x2) + ... + fp(xp) {Equation 2}

[0058] The diagrams can thus be incorporated into the respective feature-related function determination, as shown schematically in Fig. 2 by representing equation 2.

[0059] Fig. 3A shows a schematic diagram comparing a measured steering torque 300 and a torque 302 estimated according to the invention. Time t [s] is plotted on the ordinate. Torque HWT in [Nm] is plotted on the abscissa. It is clearly evident that a curve of the estimated steering torque 302 corresponds almost identically to a curve of the measured steering torque 300. The inventive estimate of the steering torque 302 is therefore almost equally good as the measured steering torque 300.

[0060] Fig. 3B shows a schematic diagram of a time course of a torque error 304 of the steering torque 302 estimated according to the invention. Time t [s] is plotted on the ordinate. The torque error, the torque HWT in [Nm], is plotted on the abscissa. It can be seen that the inventive estimation of the steering torque 302 results in only a small torque error 304, i.e., the steering torque can be estimated quite accurately. This ensures plausibility that the use of a torsion bar sensor can be dispensed with in principle, since the steering torque 302 can be estimated with sufficient accuracy.

Claims

Claims 1 . A computer-implemented method for estimating a steering torque in a steering system of a vehicle and / or a machine, the method comprising the steps: Providing (S1) a machine learning algorithm; Providing (S2) at least one input signal comprising information about at least one movement variable of the electric motor of an operating element of the steering system; and estimating (S3) the steering torque applied by an operator via the operating element by the machine learning algorithm based on the provided at least one input signal.

2. The method of claim 1, wherein the machine learning algorithm comprises an Explainable Boosting Machine, EBM, algorithm and / or a Generalized Additive Model, GAM, algorithm and / or a regression-based algorithm.

3. The method according to claim 2, wherein the GAM algorithm determines a fit function for a plurality of data points based on a plurality of cumulative spline functions, in particular using at least one regularization term.

4. The method according to claim 2 or 3, wherein the EBM algorithm is based on a GAM algorithm and is extended by a data tree structure-based approach.

5. The method according to any one of the preceding claims, wherein the machine learning algorithm is trained on the basis of training data about the at least one input signal of the electric motor of the control element of the steering system.

6. Method according to one of the preceding claims, wherein the at least one input signal comprises rotor position information and / or rotor speed information and / or rotor acceleration information and / or rotor jerk information and / or motor target torque information and / or information about a gradient of the motor target torque and / or motor actual torque information and / or information about a gradient of the motor actual torque.

7. Method according to one of the preceding claims, wherein the at least one input signal is filtered, in particular by applying a moving average and / or a low-pass filter.

8. Method according to one of the preceding claims, wherein the operating element comprises a steering wheel and / or a joystick and / or a steering lever and / or a steering switch.

9. Control unit of a steering system and / or of a vehicle, which comprises at least one processor and / or an evaluation and control device, which is configured to at least partially carry out the method according to one of the preceding claims.

10. A device (1) for estimating a steering torque in a steering system of a vehicle and / or a machine, the device (1) comprising: a provision device configured to provide a machine learning algorithm and at least one input signal comprising information about at least one movement variable of an electric motor of an operating element of the steering system; and an evaluation and computing device configured to estimate and / or approximate the steering torque applied by an operator via the operating element by applying the machine learning algorithm on the basis of the provided at least one input signal. 11 . Steering system of a vehicle and / or a machine, comprising an operating element for applying a steering torque by an operator, an electric motor coupled to the operating element; and a device according to claim 10 and / or a control device according to Claim 9.

12. A computer program comprising program code for carrying out at least parts of a method according to any one of claims 1 to 8 when the computer program is executed on a computer.

13. A computer-readable data carrier with program code of a computer program for carrying out at least parts of a method according to one of claims 1 to 8 when the computer program is executed on a computer.

Citation Information

Patent Citations

  • Method and apparatus for estimating a steering wheel torque for mechanical feedback on a steering wheel of a motor vehicle steering system

    DE102019212432B3

  • Steer-by-wire system and control method thereof

    US20200062292A1