GB2574257A

GB2574257A - Vehicle dynamics estimation method and apparatus

Info

Publication number: GB2574257A
Application number: GB1808996.1A
Authority: GB
Inventors: Nick Solomon
Original assignee: Jaguar Land Rover Ltd
Current assignee: Jaguar Land Rover Ltd
Priority date: 2018-06-01
Filing date: 2018-06-01
Publication date: 2019-12-04
Anticipated expiration: 2038-06-01
Also published as: GB201808996D0; GB2574257B

Abstract

A state estimation system (1, fig 1), for determining an estimated vehicle dynamics state EVD, comprises a processor (8) which receives a plurality of wheel speed input signals and a plurality of vehicle dynamics input signals. A measurement noise NS1 of the wheel speed input signals and/or the vehicle dynamics input signals is determined. A vehicle dynamics state PVD is predicted in dependence on the wheel speed input signals and/or the vehicle dynamics input signals. The estimated vehicle dynamics state EVD is determined in dependence on the predicted vehicle dynamics state PVD. The determination of the estimated vehicle dynamics state EVD includes applying a filter, e.g. Kalman filter 21, to the predicted vehicle dynamics state PVD. The filter 21 is controlled in dependence on the determined measurement noise NS1. The present disclosure also relates to a vehicle (1) incorporating the state estimation system (1). The present disclosure also relates to a method of estimating vehicle dynamics state EVD and a non-transitory computer-readable medium.

Description

VEHICLE DYNAMICS ESTIMATION METHOD AND APPARATUS

TECHNICAL FIELD

The present disclosure relates to a vehicle dynamics estimation method and apparatus. More particularly, but not exclusively, the present disclosure relates to a state estimation system and method for determining an estimated vehicle dynamics state.

BACKGROUND

The ability accurately to determine a current state of a vehicle, such as an automobile, can facilitate vehicle control and refinement. It is known to provide an automobile with a state estimator to estimate a current state of the vehicle utilising input signals from onboard sensors. However, known state estimators may not make use of all available sensor information. This can cause estimation inaccuracies and inconsistencies, which may potentially lead to error states in the control of onboard systems.

At least in certain embodiments, the present invention seeks to ameliorate or overcome problems associated with known state estimation systems.

SUMMARY OF THE INVENTION

Aspects of the present invention relate to a state estimation system, a vehicle, a method and a non-transitory computer-readable medium as claimed in the appended claims.

According to a further aspect of the present invention there is provided a state estimation system for determining an estimated vehicle dynamics state, the state estimation system comprising a processor configured to:

receive a plurality of wheel speed input signals;

receive a plurality of vehicle dynamics input signals;

determine a measurement noise of the wheel speed input signals and/or the vehicle dynamics input signals;

predict a vehicle dynamics state in dependence on the wheel speed input signals and/or the vehicle dynamics input signals; and determine the estimated vehicle dynamics state in dependence on the predicted vehicle dynamics state;

wherein determination of the estimated vehicle dynamics state comprises applying a filter to the predicted vehicle dynamics state, the filter being controlled in dependence on the determined measurement noise. The state estimation system may utilise input signals from a variety of sensors provided on the vehicle, including wheel speed sensors and vehicle dynamics sensors, such as accelerometers and gyroscopes. The input signals comprise measurement data output by the sensors provided on the vehicle. The noise (i.e. uncertainty) of the input signals may be modelled to enable dynamic control of the filter. The state estimation system may produce an output with reduced variance in the estimation error. At least in certain embodiments, the filter may be controlled dynamically in dependence on the determined measurement noise.

The estimated vehicle dynamics state may be determined in dependence on the predicted vehicle dynamics state and a measured vehicle dynamics state. The filter may control an extent to which the predicted vehicle dynamics state is modified in dependence on the measured vehicle dynamics state.

The processor may be configured to detect excessive noise in the input signals (for example noise outside a predefined operating range or exceeding a predetermined threshold) and to adjust the filter to refine statistical weightings applied to the predicted vehicle dynamics state. At least in certain embodiments, the state estimation system may provide improved estimation accuracy, for example during wheel slip conditions. The filter may, for example, filter the predicted vehicle dynamics state identified as being affected by the slip conditions. By monitoring fault status of one or more sensors responsible for generating the input signals, the state estimation system can allow for fault conditions, thereby providing improved tolerance to sensor error or failure.

The state estimation system may output the estimated vehicle dynamics state, for example to a stability control system, a cruise control system or an anti-lock brake system.

The gain of the filter may be adjusted in dependence on the determined measurement noise. The application of the filter may comprise applying a weighting to the predicted vehicle dynamics state. The filter may give increased weight to those estimates having a higher certainty; and less weight to those estimates having a lower certainty.

The measurement noise may be modelled in dependence on data derived from empirical testing. The testing may, for example, be performed across a dynamic range of the state estimation system.

The predicted vehicle dynamics state may be generated by a state space model. The state space model may generate the predicted vehicle dynamics state in dependence on the wheel speed input signals and/or the vehicle dynamics input signals. The determined measurement noise may represent the noise (or uncertainty) of the state space model.

The filter may be a recursive filter. The filter may be applied at least substantially in real time. The filter may be a Kalman filter (or an extended Kalman filter). The Kalman filter determines the estimated vehicle dynamics state by blending a vehicle dynamics model with the predicted vehicle dynamics state. The measurement noise associated with the input signals may be modelled. Alternatively, or in addition, model noise associated with the vehicle dynamics model may be modelled. The measurement noise and the model noise may be applied by the Kalman filter. At least in certain embodiments, the Kalman filter may reduce the variance of an estimation error (resulting from use of only the vehicle dynamics model) by fusing the probability density functions of the vehicle dynamics model and the input signals.

The measurement noise may be determined in dependence on each wheel speed input signal and/or in dependence on each vehicle dynamic input signal.

The measurement noise may be determined in dependence on a wheel slip estimation. The wheel slip estimation may be determined in dependence on the wheel speed input signals.

The measurement noise may be determined in dependence on one or more of the vehicle dynamics input signals.

The measurement noise may be determined in dependence on a sensor diagnostic signal. The vehicle dynamics input signal may be generated by one or more vehicle dynamics sensor. The sensor diagnostic signal may be output by the one or more vehicle dynamics sensor. The wheel speed input signal may be generated by one or more wheel speed sensor. The sensor diagnostic signal may be output by the one or more wheel speed sensor.

The vehicle dynamics input signals may comprise one or more of the following: a roll rate signal; a pitch rate signal; a yaw rate signal; a longitudinal acceleration signal; a lateral acceleration signal; and a longitudinal velocity signal.

Alternatively, or in addition, the vehicle dynamics signals may comprise a lateral velocity signal. The processor may be configured to estimate the lateral velocity signal in dependence on one or more of the following: a yaw rate signal, a lateral acceleration signal and a longitudinal velocity signal. The measurement noise may be determined in dependence on the lateral velocity signal.

The processor may be configured to receive the vehicle dynamics input signals from a plurality of vehicle dynamics sensors. The vehicle dynamics sensors may comprise at least one accelerometer and/or at least one gyroscope. The at least one accelerometer and/or the at least one gyroscope may be provided in an inertial measurement unit.

According to a further aspect of the present invention there is provided a vehicle comprising a state estimation system as described herein.

According to a further aspect of the present invention there is provided a method of determining an estimated vehicle dynamics state, the method comprising:

receiving a plurality of wheel speed input signals;

receiving a plurality of vehicle dynamics input signals;

determining a measurement noise of the wheel speed input signals and/or the vehicle dynamics input signals;

predicting a vehicle dynamics state in dependence on the wheel speed input signals and/or the vehicle dynamics input signals; and determining the estimated vehicle dynamics state in dependence on the predicted vehicle dynamics state;

wherein a filter is applied to the predicted vehicle dynamics state to determine the estimated vehicle dynamics state, the filter being controlled in dependence on the determined measurement noise.

The application of the filter may comprise applying a weighting to the predicted vehicle dynamics state.

The filter may be a recursive filter. The filter may be applied at least substantially in real time. The filter may be a Kalman filter.

The method may comprise determining the measurement noise in dependence on each wheel speed input signal and/or in dependence on each vehicle dynamic input signal.

Alternatively, or in addition, the method may comprise determining the measurement noise in dependence on a wheel slip estimation. The wheel slip estimation may be determined in dependence on the wheel speed input signals.

The method may comprise determining the measurement noise in dependence on one or more of the vehicle dynamics input signals.

The method may comprise determining the measurement noise in dependence on a sensor diagnostic signal. The vehicle dynamics input signal may be generated by one or more vehicle dynamics sensor. The sensor diagnostic signal may be output by the one or more vehicle dynamics sensor. The wheel speed input signal may be generated by one or more wheel speed sensor. The sensor diagnostic signal may be output by the one or more wheel speed sensor.

Alternatively, or in addition, the vehicle dynamics signals may comprise a lateral velocity signal. The method may comprise estimating the lateral velocity signal in dependence on one or more of the following: a yaw rate signal, a lateral acceleration signal and a longitudinal velocity signal. The measurement noise may be determined in dependence on the lateral velocity signal.

According to a further aspect of the present invention there is provided a non-transitory computer-readable medium having a set of instructions stored therein which, when executed, cause a processor to perform the method described herein.

Any control unit or controller described herein may suitably comprise a computational device having one or more electronic processors. The system may comprise a single control unit or electronic controller or alternatively different functions of the controller may be embodied in, or hosted in, different control units or controllers. As used herein the term “controller” or “control unit” will be understood to include both a single control unit or controller and a plurality of control units or controllers collectively operating to provide any stated control functionality. To configure a controller or control unit, a suitable set of instructions may be provided which, when executed, cause said control unit or computational device to implement the control techniques specified herein. The set of instructions may suitably be embedded in said one or more electronic processors. Alternatively, the set of instructions may be provided as software saved on one or more memory associated with said controller to be executed on said computational device. The control unit or controller may be implemented in software run on one or more processors. One or more other control unit or controller may be implemented in software run on one or more processors, optionally the same one or more processors as the first controller. Other suitable arrangements may also be used.

Within the scope of this application it is expressly intended that the various aspects, embodiments, examples and alternatives set out in the preceding paragraphs, in the claims and/or in the following description and drawings, and in particular the individual features thereof, may be taken independently or in any combination. That is, all embodiments and/or features of any embodiment can be combined in any way and/or combination, unless such features are incompatible. The applicant reserves the right to change any originally filed claim or file any new claim accordingly, including the right to amend any originally filed claim to depend from and/or incorporate any feature of any other claim although not originally claimed in that manner.

BRIEF DESCRIPTION OF THE DRAWINGS

One or more embodiments of the present invention will now be described, by way of example only, with reference to the accompanying figures, in which:

Figure 1 shows a schematic representation of a vehicle incorporating a state estimation system in accordance with an embodiment of the present invention;

Figure 2 shows a schematic representation of the state estimation system shown in Figure 1; and

Figure 3 shows a schematic representation of the operation of the state estimation system shown in Figures 1 and 2.

DETAILED DESCRIPTION

A state estimation system 1 for determining estimated vehicle dynamics states EVD of a vehicle 2 in accordance with an embodiment of the present invention will now be described with reference to the accompany figures. The estimated vehicle dynamics states EVD each represent a state of the vehicle 2 at a given time. In the present embodiment, the state estimation system 1 is configured to determine eight (8) states of the vehicle 2. In particular, the state estimation system 1 determines longitudinal velocity (v_x), lateral velocity (v_y), longitudinal acceleration (ύ_χ), lateral acceleration (ύ_χ), global pitch angle (fi_y), global roll angle (β_χ), global pitch rate (0_y), and global roll rate (β_χ).

The vehicle 2 is a road vehicle, such as an automobile, a utility vehicle or a sports utility vehicle (SUV). As shown schematically in Figure 1, the vehicle 2 comprises a body 3, an internal combustion engine 4, first and second front wheels 5-1,5-2 disposed on a front axle, and first and second rear wheels 5-3, 5-4 disposed on a rear axle. In the present embodiment, only the front wheels 5-1, 5-2 are driven (i.e., the vehicle 2 is front-wheel drive). The vehicle 2 comprises a suitable drive train (not shown) for transmitting torque generated by the internal combustion engine 4 to the front wheels 5-1,5-2. It will be understood that the state estimation system 1 described herein is applicable also to a vehicle 2 in which both the front and rear wheels 5-1, 5-2, 5-3, 5-4 are driven. The vehicle 2 is described herein with reference to a reference frame comprising a longitudinal axis X, a lateral axis Y, and a vertical axis Z, as shown in Figure 1.

The state estimation system 1 comprises an electronic control unit (ECU) 6. As shown in Figure 2, the ECU 6 comprises an electronic processor 8 and a memory 9. A set of computational instructions is stored on the memory 9 and, when executed, the computational instructions cause the processor 8 to implement the method(s) described herein. The ECU 6 is configured to receive vehicle dynamics input signals from an inertial measurement unit (IMU) 7 provided in the vehicle 2. The IMU 7 comprises an accelerometer 10 and a gyroscope 11 which, in use, measure vehicle dynamics states. The IMU 7 in the present embodiment has five (5) degrees of freedom (DOF). The accelerometer 10 has two (2) DOF and measures the following vehicle dynamics states: longitudinal acceleration and lateral acceleration In a variant, the accelerometer 10 could have three (3) DOF, for example measuring vertical acceleration (a_z(t)). The gyroscope 11 has three (3) DOF and measures the following vehicle dynamics states: local roll rate (ω_χ^), local pitch rate (ω_γ^) and local yaw rate (ω^). The local rates measured by the accelerometer 10 and the gyroscope 11 are transformed to global rates using known transformations. The IMU 7 outputs a first set of vehicle dynamics input signals S(n) representing the measured vehicle dynamics states. The vehicle dynamics input signals S(n) may be output directly to the ECU 6 or to a vehicle communication bus 12 (such as a communication CAN bus).

The IMU 7 comprises a first diagnostic module 13 configured to detect faults in the accelerometer 10 and the gyroscope 11. The first diagnostic module 13 outputs a first diagnostic signal DS1 providing an indication of the operating status of the accelerometer 10 and the gyroscope 11. If the first diagnostic module 13 detects a fault in the accelerometer 10 or the gyroscope 11, the first diagnostic signal DS1 comprises a first fault report signal (IMU_FaultDetect) for reporting the fault to the ECU 6. The first diagnostic signal DS1 may be output directly to the ECU 6 or may be output to the vehicle communication bus 12. The first fault report signal (IMU_FaultDetect) indicates which one (or more) of the dynamics sensors in the IMU7 is faulty and optionally also indicates a fault type. The first fault report signal (IMU_FaultDetect) may indicate a fault associated with a specific sensor, for example a fault associated with acceleration measurements made by the accelerometer 10 along a specific axis, or rotational measurements made by the gyroscope 11 about a particular axis.

The vehicle 2 comprises a plurality of wheel speed sensors 14-n, each wheel speed sensor 14-n being associated with a respective one of said wheels 5-n. The wheel speed sensors 14n publish wheel speed input signals WS(n) to the vehicle communication bus 12 for access by the state estimation system 1. The wheel speed input signals WS(n) comprise a measured wheel speed of each wheel 5-n (WhlSpd_FL, WhlSpd_FR, WhlSpd_RL, WhlSpd_RR). The longitudinal velocity (v_x) of the vehicle 2 is determined from the measured wheels speeds in conventional manner. A slip estimator 15 is provided for estimating wheel slip in dependence on the wheel speed input signals WS(n). The slip estimator 15 receives the measured wheel speed of each wheel 5-n and estimates a wheel slip of each wheel 5-n using known techniques. The slip estimator 15 may, for example, compare the rotational speeds of the (driven) front wheels 5-1,5-2 with the rotational speed of the (undriven) rear wheels 5-3, 5-4 to estimate wheel slip of the front wheels 5-1,5-2. Other known techniques to estimate wheel slip include comparing the rate of change of the wheel speed (i.e. angular acceleration) with a predetermined threshold value. For example, if the wheel acceleration exceeds a predefined threshold, is identified. Alternatively, or in addition, the measured rotational speed of the wheels 5-n on opposing sides of the vehicle 2 may be compared with each other. If the difference in the rotational speed of the wheels on the left and right sides exceeds a predefined threshold, wheel slip is identified. The slip estimator 15 outputs a wheel speed noise detect signal (WhlSpd_NoiseDetect) providing an indication of the statistical noise in the measured wheel speed of each wheel 5-n (WhlSpd_FL, WhlSpd_FR, WhlSpd_RL, WhlSpd_RR). The wheel speed noise detect signal (WhlSpd_NoiseDetect) may be generated if noise is detected in the wheel speed measured by a wheel speed sensor 14-n. The wheel speed noise detect signal (WhlSpd_NoiseDetect) may be generated if the detected noise is outside a predefined operating range or greater than or less than a predefined threshold. The wheel speed noise detect signal (WhlSpd_NoiseDetect) may indicate an estimated amount of wheel slip of each wheel 5-n; or may comprise a flag to indicate whether or not wheel slip is detected at each wheel 5-n. The wheel speed noise detect signal (WhlSpd_NoiseDetect) is output to the state estimation system 1. In a variant, the slip estimator 15 may form part of the state estimation system 1.

A second diagnostic module 16 is configured to detect faults in the wheel speed sensors 14n. The second diagnostic module 16 may monitor the wheel speed sensors 14-n or may be configured to receive an error signal from the wheel speed sensors 14-n. The second diagnostic module 16 outputs a second diagnostic signal DS2 providing an indication of the operating status of each wheel speed sensor 14-n. If the second diagnostic module 16 detects a fault in one of the wheel speed sensors 14-n, the second diagnostic signal DS2 comprises a second fault report signal (WhlSpd_FaultDetect) identifying the fault. The second fault report signal (WhlSpd_FaultDetect) indicates which one (or more) of the wheel speed sensors 14-n is faulty and optionally also indicates a fault type. The second diagnostic signal DS2 may be output directly to the ECU 6 or may be output to the vehicle communication bus 12.

In the present embodiment, the lateral velocity (v_y) of the vehicle 2 is determined by a lateral velocity estimator 17. The lateral velocity estimator 17 estimates the lateral velocity (v_y) in dependence on the following inputs: the longitudinal velocity (v_x) determined from the wheel speed input signals WS(n); the local yaw rate (ω_ζ^) measured by the gyroscope 11; and the local lateral acceleration (a_y(t)) measured by the accelerometer 10. The calculation of the lateral velocity (v_y) is known in the art. The lateral velocity estimator 17 outputs a second set of the vehicle dynamics input signals S(n) comprising the lateral velocity (v_y). In the illustrated arrangement, the lateral velocity estimator 17 is a separate module provided in the vehicle 2. In a variant, the lateral velocity estimator 17 may be incorporated into the state estimation system 1.

A schematic representation of the state estimation system 1 is shown in Figure 3. The processor 8 is configured to implement a state space model (denoted generally by the reference numeral 20). As described herein, the state space model comprises a state prediction module 20A and a state measurement module 20B. The state prediction module 20A outputs a set of predicted vehicle dynamics states PVD; and the state measurement module 20A outputs a set of measured vehicle dynamics states MVD.

The processor 8 receives the wheel speed input signals WS(n) output by the wheel speed sensors 14-n; the first set of the vehicle dynamics input signals S(n) output by the IMU 7; and the second set of the vehicle dynamics input signals S(n) output by the lateral velocity estimator 17. The wheel speed input signals WS(n) comprise the measured wheel speed of each wheel 5-n (WhlSpd_FL, WhlSpd_FR, WhlSpd_RL, WhlSpd_RR). The first set of the vehicle dynamics input signals S(n) comprises the global roll rate (β_Χ(ρ), the global pitch rate (0_y(_t)), the global yaw rate (0_z(t)), the longitudinal acceleration (ύ_χ^) and the lateral acceleration The longitudinal velocity (v_x) and the longitudinal acceleration (ύ^) are determined from the measured wheel speed of each wheel 5-n (WhlSpd_FL, WhlSpd_FR, WhlSpd_RL, WhlSpd_RR). The second set of the vehicle dynamics input signals S(n) comprises the lateral velocity (v_y). In the present embodiment the lateral velocity (v_y) is modelled by the lateral velocity estimator 17 in dependence on the longitudinal velocity (v_x). The lateral acceleration may, for example, be determined in dependence on the global yaw rate (0_z(t)), the lateral acceleration and the lateral velocity (v_y).

The processor 8 implements a Kalman filter 21 (also known as a linear quadratic estimation algorithm) dynamically to filter predicted vehicle dynamics states PVD generated by a state space model. The Kalman filter 21 is controlled in dependence on a measurement noise signal NS1 generated by a noise model 22. In the present embodiment, the noise model 22 is implemented by the processor 8, but it could be a separate module configured to output the measurement noise signal NS1 to the ECU6. The operation of the noise model 22 will now be described. The Kalman filter 21 has a gain K which determines the extent to which the state prediction matrix x is corrected by the state measurement matrix y. As described herein, the gain K is determined by the noise model 22 in dependence on the wheel speed noise detect signal (WhlSpd_NoiseDetect), the first fault report signal (IMU_FaultDetect) and the second fault report signal (WhlSpd_FaultDetect).

The noise model 22 receives the first diagnostic signal DS1, the second diagnostic signal DS2 and the wheel speed noise detect signal (WhlSpd_NoiseDetect). The noise model 22 checks for the presence or absence of the first fault report signal (IMU_FaultDetect) in the first diagnostic signal DS1; and for the presence or absence of the second fault report signal (WhlSpd_FaultDetect) in the second diagnostic signal DS2. The noise model 22 may thereby identify any faults present in the measurements received from the accelerometer 10 and the gyroscope 11. The noise model 22 may also identify measurement noise in dependence on the wheel speed noise detect signal (WhlSpd_NoiseDetect). The noise model 22 outputs a measurement noise (i.e. a measurement uncertainty) in dependence on the first fault report signal (IMU_FaultDetect), the second fault report signal (WhlSpd_FaultDetect) and the wheel speed noise detect signal (WhlSpd_NoiseDetect). Alternatively, or in addition, the noise model 22 may use the determined lateral velocity (v_y) of the vehicle 2 as an input for the noise modelling. The modelled measurement noise changes dynamically with respect to time. The modelled measurement noise determined by the noise model 22 is output to the Kalman filter 21 to determine the estimated vehicle dynamics states EVD.

The measured signals in the state measurement matrix y have an associated noise value e(t) which represents a measurement uncertainty. As described herein, the noise values e(t) represent a covariance matrix of the measurement noise Rv. The first fault report signal IMU_FaultDetect, the second fault report signal WhlSpd_FaultDetect and the wheel speed noise detect signal WhlSpd_NoiseDetect adjust the noise (i.e. the uncertainty or, in statistical terms, the variance) of each measurement signal. The measurement noise covariance matrix Rv is used to determine a gain K for the Kalman filter 21. The determined gain K is applied during a correction process to change an extent to which the predicted vehicle dynamics matrix PVD generated by the state prediction module 20A is corrected by the measured state vehicle dynamics matric MVD generated by the state prediction module 20B. The gain K may be understood as applying a “weighting” to the predicted vehicle dynamics matrix PVD. As described herein, the weighting is implemented by a noise correction matrix Rv Corr. The gain K applies a lower weighting to a measurement signal which is identified as having a high noise during the correction step.

In use, the processor 8 first implements a state space model to generate the predicted vehicle dynamics states PVD. The state space model utilises the current state input signals, including: wheel speeds (WhlSpd_FL, WhlSpd_FR, WhlSpd_RL, WhlSpd_RR), longitudinal velocity (v_x), lateral velocity (v_y), longitudinal acceleration (ύ_χ^), lateral acceleration global roll angle (θ_χ), global pitch angle (6_y); global roll rate (θ_χ^); and global pitch rate (θ_γ^) to generate the predicted vehicle dynamics states PVD. In the present embodiment, the state prediction module 20A applies the following state space model vehicle dynamics states PVD:

to generate the predicted

			Ax
1......	0	r	0	0	0	0	0'		it	0	0	0
	1	0	T	0	0	0	0		1	δ	0	0
11	0	1	0	0	0	0	0	ΐί' / 1 i ·ρ >·. ., v	lb	0	0	δ
iil	0	0	1	0	0	0	0	I s	0	1	0	0
111	0	0	0	1	0	Τ-	0	j*\|	it	0	0	0
ill	0	0	0	0	1	Ο	T	1		0	0	0
lit	δ	0	0	0	0	1	0		o	δ	1.	0
1	0	0	0	0	0	0	1-	\|	o	0	0	1

£) J

Whereby: x is the predicted state;

A is a state transformation matrix;

G is a noise application matrix;

w is a random walk for each signal;

t represents time;

y is the measurement vector;

v_x is longitudinal velocity;

v_y is lateral velocity;

v_x is global longitudinal acceleration;

v_y is global lateral acceleration;

θ_χ is the global roll angle;

6_y is global pitch angle;

θ_χ is global roll rate; and e_y is global pitch rate.

The random walk w is applied to track variables with uncertainty. The random walk w adds a random noise value to the highest derivatives in the state vector x. The noise application matrix G defines application of the random walk w to the states defined in the state vector x. In the present embodiment, the random walk w is applied to all the rates in the state vector x, i.e. longitudinal acceleration (ύ_χ), lateral acceleration (ύ_χ), global pitch rate (0_y), and global roll rate (θ_χ). The measurement noise (e) is described herein. The local longitudinal acceleration a_xs and the local lateral acceleration a_xs are measured by the IMU 7.

The measurement of longitudinal velocity (u_x(t)) is taken from the wheel speed sensors 14-n; and the lateral velocity (u_y(t)) is calculated, as described herein. The accelerometer measurements are corrected for centripetal acceleration and gravity contamination. The noise model is added giving the following equations:

The measurements of pitch and roll angular velocities are converted from a local to global coordinate system and, again, the noise model added:

() ™ ) : ^: G j Τ ^:®(ϊ) <’A^sto ” .^^: - ferir ^ω?(ί > '1“

In the present embodiment, the state measurement module 20B applies the following state space model to generate the measured vehicle dynamics states MVD:

^x(t)'	..	0	0	0	0	0	0
Vy{t)	0	1	0	0	0	0	0
^XSiT)	o		1	0	*,9	0	e
		0	0	1	0	9	0
ii-L-it)	0	0	(}	0	”^ωζ£ί)	0	0
Λ{0-	- 0:	0	0	0	0		1

1.


^vyit)		^t,·
		‘ Vi £1

0_y<t,	+	. . z>


^y(r)
j£j
«TP

Whereby:

y is the measured state; t represents time; v_x is longitudinal velocity; v_y is lateral velocity; v_x is global longitudinal acceleration; v_y is global lateral acceleration; a_xs is longitudinal acceleration; a_ys is lateral acceleration; ω_χ is local roll rate; co_y is local pitch rate; θ_χ is the global roll angle; θ_γ is global pitch angle; θ_χ is global roll rate; 6_y global pitch rate; g is gravitational acceleration; C is a measurement transformation matrix; and

e is measurement noise.

The Kalman filter 21 is implemented as follows:

Stote prediction step:

Measi/re/went correction step:

/Q_o ™ ...j) C|Rp. + iT A_t;f ¹ ™ Αί :ί-· } ~ ^ij ' tos:;.-rJ μ - κ

Whereby K is the Kalman gain; and

Rw is the error covariance matrix of the process noise.

The process noise covariance matrix Rw is defined by the random walk w as follows:

Rw ” E{w_jn ^s w_ii:J

The process noise variance is the standard deviation squared.

The measurement noise covariance matrix Rv is defined by the noise model:

^e«ws(i) p P ''^νϊΐ'ί ’ £tjf yf

-

	0	0	0	0 0
		0	0	0 0
0			0	0 0
0	0			0 0
0 0	0 0	0 0	0	0

- 0	0	0	0	0 J

The measurement noise variance σ]^ is the standard deviation squared.

The longitudinal velocity (u_x), lateral velocity (u_y), longitudinal acceleration (ΰ_χ), lateral acceleration (ϋ_χ), global pitch angle (0_y), global roll angle (0_X), global pitch rate (0_y), and global roll rate (0_X) can be predicted. The processor 8 then applies the Kalman filter 21 to update the predicted vehicle dynamics states PVD in dependence on the modelled measurement noise. The Kalman filter 21 corrects the predicted vehicle dynamics states PVD towards measured vehicle dynamics states MVD in dependence on the gain K. The predicted vehicle dynamics states PVD are updated using a weighted average determined in dependence on the modelled measurement noise which provides an indication of the certainty of the predicted vehicle dynamics states PVD. The amount the predicted vehicle dynamics states PVD are corrected towards the measured vehicle dynamics states MVD is determined by the gain K of the Kalman filter 21. This is performed recursively, at every time step, so the weighting applied to the predicted vehicle dynamics states PVD changes with time. The recursive implementation of the Kalman filter 21 provided a dynamic, fault tolerant model. The gain K of the Kalman filter 21 is determined by the measurement uncertainty Rv. The Kalman filter 21 gives increased weight to those predicted vehicle dynamics states PVD having a higher certainty; and less weight to those predicted vehicle dynamics states PVD having a lower certainty. The estimated vehicle dynamics states EVD are generated by applying the Kalman filter 21 to the predicted vehicle dynamics states PVD. The Kalman gain K is controlled in dependence on the modelled measurement noise. A noise detection event, such as detection of wheel slip, is used to increase the variance of the noise from the wheel speed measurement. This in turn reduces the gain K giving greater emphasis to the predicted vehicle dynamics states PVD and less emphasis to the measured vehicle dynamics states MVD.

The processor 8 outputs the estimated vehicle dynamics states EVD to a controller 24, for example a stability control system. In particular, the processor 8 outputs the following estimated vehicle dynamic parameters EVD, including the longitudinal velocity (v_x), lateral velocity (v_y), longitudinal acceleration (ύ_χ), lateral acceleration (ύ_χ), pitch angle (0_y), roll angle (β_χ), pitch rate (0_y), and roll rate (β_χ). It will be understood that the estimated vehicle dynamics states EVD may be used by other vehicle systems, for example a stability control system.

The state estimation system 1 according the present embodiment may provide improved estimation accuracy, for example during wheel slip conditions. The Kalman filter 21 is applicable to filter the predicted vehicle dynamics state identified as being affected by the slip conditions. Furthermore, by monitoring the fault status, the noise model 22 can respond to any faults identified in the sensors, such as the wheel speed sensors 14-n. At least in certain embodiments, the state estimation system 1 may show improved tolerance to sensor failure.

It will be appreciated that various modifications may be made to the embodiment(s) described herein without departing from the scope of the appended claims.

Claims

1. A state estimation system for determining an estimated vehicle dynamics state, the state estimation system comprising a processor configured to:

receive a plurality of wheel speed input signals;

receive a plurality of vehicle dynamics input signals;

wherein determination of the estimated vehicle dynamics state comprises applying a filter to the predicted vehicle dynamics state, the filter being controlled in dependence on the determined measurement noise.

2. A state estimation system as claimed in claim 1, wherein the estimated vehicle dynamics state is determined in dependence on the predicted vehicle dynamics state and a measured vehicle dynamics state.

3. A state estimation system as claimed in claim 2, wherein the filter controls an extent to which the predicted vehicle dynamics state is modified in dependence on the measured vehicle dynamics state.

4. A state estimation system as claimed in any one of claims 1,2 or 3, wherein applying the filter comprises applying a weighting to the predicted vehicle dynamics state.

5. A state estimation system as claimed in any one of the preceding claims, wherein the filter is a Kalman filter.

6. A state estimation system as claimed in any one of the preceding claims, wherein the measurement noise is determined in dependence on each wheel speed input signal and/or in dependence on each vehicle dynamic input signal.

7. A state estimation system as claimed in any one of the preceding claims, wherein the measurement noise is determined in dependence on awheel slip estimation.

8. A state estimation system as claimed in any one of the preceding claims, wherein the measurement noise is determined in dependence on one or more of the vehicle dynamics input signals.

9. A state estimation system as claimed in any one of the preceding claims, wherein the measurement noise is determined in dependence on a sensor diagnostic signal.

10. A state estimation system as claimed in any one of the preceding claims, wherein the vehicle dynamics input signals comprise one or more of the following: a roll rate signal; a pitch rate signal; a yaw rate signal; a longitudinal acceleration signal; a lateral acceleration signal; and a longitudinal velocity signal.

11. A state estimation system as claimed in any one of the preceding claims, wherein the vehicle dynamics signals comprise a lateral velocity signal.

12. A state estimation system as claimed in any one of the preceding claims, wherein the processor is configured to receive the vehicle dynamics input signals from a plurality of vehicle dynamics sensors.

13. A state estimation system as claimed in claim 12, wherein the vehicle dynamics sensors comprise at least one accelerometer and/or at least one gyroscope.

14. A vehicle comprising a state estimation system as claimed in any one of the preceding claims.

15. A method of determining an estimated vehicle dynamics state, the method comprising:

receiving a plurality of wheel speed input signals;

receiving a plurality of vehicle dynamics input signals;

16. A method as claimed in claim 15, wherein the estimated vehicle dynamics state is determined in dependence on the predicted vehicle dynamics state and a measured vehicle dynamics state.

17. A method as claimed in claim 16, wherein the filter controls an extent to which the predicted vehicle dynamics state is modified in dependence on a measured vehicle dynamics state.

18. A method as claimed in any one of claims 15, 16 or 17, wherein applying the filter comprises applying a weighting to the predicted vehicle dynamics state.

19. A method as claimed in any one of claims 15 to 18, wherein the filter is a Kalman filter.

20. A method as claimed in any one of claims 15 to 19 comprising determining the measurement noise in dependence on each wheel speed input signal and/or in dependence on each vehicle dynamic input signal.

21. A method as claimed in any one of claims 15 to 20 comprising determining the measurement noise in dependence on awheel slip estimation.

22. A method as claimed in any one of claims 15 to 21 comprising determining the measurement noise in dependence on one or more of the vehicle dynamics input signals.

23. A method as claimed in any one of claims 15 to 22 comprising determining the measurement noise in dependence on a sensor diagnostic signal.

24. A method as claimed in any one of claims 15 to 23, wherein the vehicle dynamics input signals comprise one or more of the following: a roll rate signal; a pitch rate signal; a yaw rate signal; a longitudinal acceleration signal; a lateral acceleration signal; a longitudinal velocity signal; and a lateral velocity signal

25. A non-transitory computer-readable medium having a set of instructions stored therein which, when executed, cause a processor to perform the method claimed in any one of claims 15 to 24.