Method for correcting the delay of a dynamic simulator

The method addresses simulator delay compensation by modifying the reference signal with predictive algorithms and filtering techniques to advance the phase uniformly, effectively correcting delays and reducing simulator sickness.

FR3117255B1Active Publication Date: 2024-11-29RENAULT SA +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
FR2020012842
Authority / Receiving Office
FR · FR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2020-12-08
Publication Date
2024-11-29
Estimated Expiration
2040-12-08

AI Technical Summary

Technical Problem

Existing simulator delay compensation methods are limited to a narrow frequency band and cause signal distortion and simulator sickness due to non-linear phase shifts and excessive reduction of low-frequency amplitudes.

Method used

A method involving predictive algorithms to modify the reference signal by adding predictive signals, filtering low frequencies, and applying a Padé formula to advance the phase uniformly, using a combination of Fourier transforms and digital filters to correct simulator delays.

Benefits of technology

The method effectively corrects simulator delays in real-time, broadening the frequency band and reducing simulator sickness by minimizing signal distortion and maintaining phase linearity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000016_0000
    Figure 00000016_0000
  • Figure 00000016_0001
    Figure 00000016_0001
  • Figure 00000016_0002
    Figure 00000016_0002
Patent Text Reader

Abstract

Method for correcting the delay of an automotive driving simulator for the rendering of motion of a dynamic vehicle model piloted by a driver located in said simulator comprising a simulation platform capable of moving in a substantially flat workspace along a path calculated in real time and controlled following a path of the vehicle in a virtual space, said commands being optimized by a real-time motion rendering algorithm, said platform is controlled by a central unit running said method which receives a reference signal and is capable of sending said modified signal to a control algorithm capable of generating commands to be passed to the animation means of the simulation platform.characterized in that it comprises the following consecutive steps: - a first step of modification (20) of the reference signal by adding predictive signals; - a second step, following the first step, of decomposition (30) of the modified reference signal into a combination of frequencies; - a third step, following the second step, of frequency filtering (40) of the signal comprising a frequency modulation profile; - a fourth step, following the third step, of signal recomposition (50). Figure for the abbreviation: Figure 3.
Need to check novelty before this filing date? Find Prior Art

Description

Title of the invention: Method for correcting the delay of a dynamic simulator Technical field of the invention

[0001] The present invention relates to a car driving simulator.

[0002] The present invention relates to a method for correcting the delay of a simulator dynamic.

[0003] The purpose of the method of correcting the delay of a driving simulator is to make the sensation of movement perceived in the simulator as natural as possible by reducing the additional latency of the simulator. State of the art

[0004] In the field of automobile driving simulations, it is possible to have a simulation platform or simulator capable of moving in an environment or workspace of reduced or limited dimensions.

[0005] The simulator consists of a cockpit, a mobile platform and a display screen. A driver's station is installed in the cockpit.

[0006] The simulator is mobile in the workspace which is for example a closed room with a flat surface and the mobile platform can move in the four corners of said flat surface of the room. However, the reduced space only allows sensations of limited duration or limited amplitude to be reproduced. To overcome this constraint of reduced dimensions, anticipation of the movements of the mobile platform, both linear and rotational, is necessary to ensure acceleration levels that can be felt by the users. It is therefore necessary to take into account the position as well as the speed and acceleration values ​​of the platform.

[0007] To compensate for the dimensional constraints of the reduced mobility space in which the platform moves, it is also known to associate gravity effects, rotational movements of the cabin inclination, with the plane movements of said platform. Indeed, the human brain is sensitive to the variation of acceleration, and physiological effects can be obtained by the use of devices to compensate for the limited movements along a plane of the platform.

[0008] Certain devices make it possible in particular to restore long-lasting effects. It is thus known to carry out a cab rearing with, for example, an angular speed lower than a threshold of human perception to give an impression of linear acceleration supported by the gravity effect which acts in the vestibular system as well as that of the weight of the body on the back of the driver's seat.

[0009] Simulators must produce perceptions of movements close to the real physical vehicle especially for a driver at the wheel with different driving movements such as turns, changing lanes.

[0010] A disadvantage of large simulators is their latency time due in particular to their weight and / or performance. They are therefore likely to generate a delay which is far from negligible and which can be felt by the driver.

[0011] Delay compensation algorithms have been developed.

[0012] In a known manner, a pure delay, noted can be represented in the field of signal processing by an exponential function in the form

[0013] [Math.l] exp (— ç.s)

[0014] of which s is the variable of the Laplace transform.

[0015] The response of a delayed signal is determined by:

[0016] [Math.2] F (s) = exp .X(s)

[0017] where X(s) is the signal input of a system expressed in the Laplace domain and Y(s) is the output or response of the delay system.

[0018] If we want to keep the response Y(s) without delay, we must then correct the phase of the input signal X(s) by the inverse of the pure delay function, i.e.:

[0019] [Math.3] X- (s) = exp s) , X(s)

[0020] X*(s) is the signal response with a phase advance which is the product or the operation of the convolution between the Laplace transform of the signal impulse and that of the inverse of the pure delay.

[0021] The time signal x*(t) corresponding to X*(s) can be deduced via a prediction of the signals at the time horizon,d, for example by a Taylor series development:

[0022] [Math.4] 4- rd) = 4- 4- ^"(0' $1 / 2 T "■

[0023] In this way, we can find the different works published in the literature. We note that in 2005, NASA presented several algorithms to compensate for the delay of the flight simulator. The generalized prediction algorithm has the expression:

[0024] [Math.5]

[0025] Note that Taylor's formula could be part of this generalized expression.

[0026] Another published method presents a delay compensation based on a first or second order approach of Taylor series combined with the MPC technique (Model Predictive Control, "Performance identification and compensation of simulator motion cueing delays", DSC2010, Z. Fang et al., "An efficient MPC based motion cueing delay compensation algorithm for driving simulator", DSC2015, Z. Fang et al.) with an order 1 or 2 of the Taylor series development which corresponds respectively to the Padé (1,0) or Padé (2,0) formula, known in the scientific literature.

[0027] Cruden and the University of Stuttgart have also presented a prediction method of the type of Taylor series expansion to order 2.

[0028] In 2020, we presented a method for compensating for simulator delay by reducing low-frequency components of the signal ("Motion cueing algorithms and objective criteria investigation for reproducing high lateral dynamic motion perception in driving simulator", Z. FANG et al. DSC2020).

[0029] Broadly speaking, there are two methods for advancing the phase of a reference or base signal by doing

[0030] -a prediction of the signals

[0031] -a reduction of low frequency components.

[0032] A disadvantage is that all prediction methods published in the literature are only valid for a certain frequency band which is quite limited because of the moduli of the high frequency signals which are non-unitary and amplified exponentially.

[0033] The disadvantage for the method of reducing low frequency components is that:

[0034] -on the one hand, the real-time filter presents too significant distortion of the signals because of the phase non-linearity of the filter, that is to say that the signal advance times depend strongly on the signal frequency.

[0035] -and on the other hand, the amplitudes of the low-frequency signals could be too small to obtain a sufficient or necessary phase advance. Consequently, the reference signal thus distorted could cause the phenomenon of simulator sickness.

[0036] There is therefore a need for a method for correcting delays for a simulator or a driving simulation platform equipped with an algorithm for restoring dynamic movement of a vehicle driven by a driver who is inside a cabin of said simulator, which is composed of a simulation platform capable of moving in a substantially flat and limited workspace. Presentation of the invention

[0037] The subject of the invention is a method for correcting the simulator delay of car driving for a movement restitution of a dynamic model of a vehicle driven by a driver who is in said simulator comprising a simulation platform capable of moving in a substantially planar workspace according to an optimized and controlled trajectory following a route of the vehicle in a virtual environment, said commands being optimized by a real-time algorithm for the movement restitution, said platform is controlled by a central unit carrying out said method which receives an input reference signal and is capable of sending said modified signal to a command algorithm capable of generating commands to be passed to the animation means of the simulation platform.

[0038] Characterized in that it comprises the following consecutive steps:

[0039] - a first step of modifying the frame of the reference signal by adding predictive signals to a part of said frame formed from the most recent last points of the reference signal.

[0040] - a second step following the first step of identifying the reference signal modified in frequency distribution.

[0041] - a third step following the second step of filtering the frequencies of the signal comprising a frequency modulation profile,

[0042] - a fourth step following the third step of recomposing the input signal.

[0043] Advantageously, the method for correcting the delays of the automobile simulation device proposes modifying the reference signal to correct the delays of said device, in particular by using a predictive algorithm and modulating the frequencies of the input signal to make a phase advance of the signal in a homogeneous manner in terms of time relative to the origin of the signal.

[0044] According to other characteristics of the invention:

[0045] - the frame of the reference signal taken into account is reduced to the last points of said frame representing a maximum sequence of 5 seconds.

[0046] Advantageously, the invention retains and takes into account at each instant t the last N points of the frame forming the input reference signal, the number N can vary according to powers for example of 2 from 256 to 2048 forming a sequence of a few seconds, of the order of 1 to 5 seconds of the current signal 'instant t. The reduced reference frame is modified by adding predictive signals. Thus, the frame of the reference signal comprises characteristics as close as possible to the instant t.

[0047] -the first step involves adding an additional sub-frame of predictive signals arranged at the start of the frame of the reference signal.

[0048] Advantageously, the additional sub-frame makes it possible to improve the resolution of the frequencies of the signal spectrum. Indeed, said frame makes it possible to form a periodic sequence of the reference signal.

[0049] -the additional subframe comprises a sequence of zero values.

[0050] Advantageously, the additional subframe is formed from a sequence of values ​​to ensure the continuity of the periodic signals.

[0051] -the first step comprises a generation of a frame which is composed of the reduced reference frame and an additional symmetrical part of said frame arranged at the start of the reduced reference frame.

[0052] Advantageously, the additional sub-frame is formed from a symmetrical frame of the reference signal so that the amplitudes of the spectra of the additional sub-frame are very close to those of the reference signal, the additional sub-frame is formed from a symmetrical sequence of the reference signal with respect to a time axis. The reference signal is then represented in a reference frame of frequency / amplitude axes. Thus, a periodicity of the spectra of the signal is obtained for its subsequent processing.

[0053] -the additional subframe comprises a sequence of predictive signals resulting from the transformation of the reference signal.

[0054] Advantageously, the method comprises a phase of advance transformation of the reference signal with a convolution of the inverted delay function.

[0055] -the frame of predictive signals is added at the start of the additional subframe.

[0056] Advantageously, to have a modified and periodic reference signal, the predicative signal frame is added upstream of the additional subframe, to take into account the periodicity, the continuity of the signal as well as the operation of phase advancement.

[0057] -The third step includes a modulation of the amplitudes of the frequency spectra of the signal.

[0058] Advantageously, the third step makes it possible to modulate the amplitudes of the decomposed reference signal as a function of the frequency in order to retain a desired frequency band.

[0059] As is known, the frequency spectrum of a time domain of a signal is the representation of this signal in the frequency domain.

[0060] -The third step comprises a modulation of the amplitudes of the spectra as a function of the frequencies of the signal according to a modulation profile.

[0061] - The third step includes a modulation of the amplitudes of the combination of the frequencies according to a modulation spectrum according to the frequency.Advantageously, a modulation spectrum of the amplitudes according to the frequencies is proposed to be applied to the modified and decomposed reference signal to filter the frequency band of the decomposed reference signal.

[0062] Cleverly, we can increase the advance of the recomposed signal with respect to the reference signal by eliminating the low frequencies or by reducing the low frequency weighting values ​​in the modulation spectrum.

[0063] -the method allows the parameterization of the delay correction

[0064] Advantageously, a recalibration is carried out between the first reference signal and the second recomposed signal which has a phase or time advance compared to the first signal, for example by intercorrelation which is a simple and rapid method of checking the phase shift between the two signals.

[0065] Depending on the results of said verification, the delay correction can be modified in a simple manner by parameterization.

[0066] - the delay correction setting includes a modification of the sequence predictive signals via modification of the delay time to be corrected

[0067] Advantageously, a modification of the sequence of predictive signals makes it possible to simply correct the delay of the simulator by modifying the modeling parameters of the predictive signals.

[0068] - the delay correction setting includes a modification of the profile of the modulation of the amplitudes of the frequency spectra.

[0069] - the delay correction setting includes a modification of the mo dulation of the amplitudes of the frequency components of the modified reference signal.

[0070] Advantageously, it is also possible to act on the modulation of the amplitudes of the frequency spectra which thus makes it possible to reduce the low frequency signals. Detailed description of the invention

[0071] Other characteristics and advantages of the invention will appear on reading the following description of particular embodiments of the invention given as non-limiting examples and represented in the appended drawings, in which:

[0072] [Fig-1] is a schematic view of a car driving simulation device with commands generated by a real-time system.

[0073] [Fig.2] is a schematic view of the signal processing by the simulator delay correction method.

[0074] [Fig.3] is a schematic view of the operating flowchart of the delay correction method.

[0075] In the following description, identical reference numerals designate identical parts or parts having similar functions.

[0076] As is known, car driving simulators are of two types:

[0077] -a first type of simulator without movement and

[0078] -a second type of simulator with a mobile platform and cabin, which is the object of our invention.

[0079] As shown in [Fig.l], a driving simulator or simulation device 50 may comprise a cabin 51 adapted to accommodate a driver and held by a hexapod comprising extendable arms or actuators 52, each of the arms being fixed by one end to a platform 53. Said platform is movable in a work space 54 here by sliding on rails 55.

[0080] Said workspace 54 has a substantially flat surface. It can be formed by a room with dimensional limits. The cabin 51 commonly called cockpit, is fixed on the upper part of the hexapod which is held by the extendable arms 56 and actuators 52. The mobile platform 53 or upper base of hexapod can move relative to a horizontal plane substantially parallel to the floor of the workspace 54, to produce linear movements. The extendable arms are capable of producing movements and rotations of the cabin relative to said horizontal plane. The mobile platform 53 with the cabin 51 are movable thanks to activation or animation means 57 which are controlled by a central unit (not shown). Said extendable arms 52 as well as a means of moving the mobile platform 53 are part of the animation means 57 of the mobile platform 53 with the cabin 51.Said central unit is capable of defining commands which are then addressed to said means to physically move the mobile platform and the cabin.

[0081] The driver is placed in a virtual environment and drives a digital vehicle in said virtual driving environment. The images are projected in front of the driver's vision to provide a virtual driving environment. In said environment, the driver drives the vehicle according to a predefined scenario, for example free driving on a road.

[0082] The driver generates driving commands which are addressed to the central unit throughout the journey.

[0083] The central unit runs a command algorithm 11 to generate commands to be passed to the animation means of the simulation platform 57 following the driving commands. Said commands for the animation means are generated per time interval or time cycle of between one and ten milliseconds.

[0084] Said algorithm 11 comprises a calculation program which is a real-time program capable of controlling or piloting the movements of said simulation platform 53. It is therefore necessary that the results of the calculations are delivered at a speed adapted to the movements of said platform 53.

[0085] The central unit also controls the display of images in front of the driver and collects driving commands issued by the driver depending, for example, on the trajectory of the road to be followed. The driver steers the virtual vehicle along a route including straight lines and bends.

[0086] Simulators must produce perceptions of movements close to the real physical vehicle, especially for a driver at the wheel with different driving maneuvers such as turns, changing lanes.

[0087] High-performance simulators are specified to operate with a cycle time of 1 ms and a dynamic response time of around 20 ms. To enable the simulator to approximate the inertial forces perceived in a vehicle, a motion restitution algorithm is implemented.

[0088] The simulator's motion perception must match that of a physical vehicle when the driver performs maneuvers such as slalom, double lane changes or vehicle acceleration. Unfortunately, all dynamic simulators have a more or less significant latency depending on the performance and weight of the simulator.

[0089] As a result, large, very high-performance simulators are likely to generate a far from negligible delay as a consequence of their significant loads.

[0090] The object of our invention is to modify the driving command signal before it is taken into account by the central unit to generate the commands to the animation means of the simulator.

[0091] According to [Fig.2], the initial movement signal of the driving or reference vehicle 12 passes through a computer which can be the central unit, which will carry out a correction method 10 of the delays of the simulator, the subject of our invention. The modified reference signal 12', the method of the invention, will be entered into the central unit for the generation of said commands 11.

[0092] According to the invention, the method 10 proposes to filter the low frequencies of the signal and to keep only a small part of the signal at low frequencies by using a conventional second-order filter. It is recalled that a high-pass filter, called HP filter or in English "high-pass" of second order presents a phase shift of the signals in particular for the low-frequency components.

[0093] We can therefore use this phase shift of the signals to compensate for the delay of the simulator.

[0094] In a known manner, a reconstructed signal can be defined by:

[0095] [Math.6] 54 = - JW) + ^2-Thp

[0096] Where:

[0097] x is the original signal

[0098] yHp is the response to the HP high-pass filter of the original reference signal x 12.

[0099] If ki is equal to k2, i.e. ki = k2jyr = kp x , a gain ki is applied to the signal original. The resulting signal does not change phase relative to the original signal X .

[0100] If ki is less than k2, that is, ki < k2, then the resultant yr is ahead in phase with respect to x.

[0101] It is therefore sufficient to properly dimension ki and k2 so that the phase advance compensates for the simulator delay.

[0102] The method for correcting the simulator delays comprises a first step 20 during which the reference signal 12 is modified with the addition of an additional sub-frame 19 comprising a sequence of prediction signals 16 or phase advance

[0103] The reference signal may extend over a time sequence of several seconds. According to the invention, the reference signal 12 taken into account is a reduced part of the frame of the reference signal. Said reduced frame is formed of the last N points of the current signal before the instant t. The number N is a multiple of 2 from 256 to 2048 for example.

[0104] Said predictive signal advance sequence 16 is defined in the following part.

[0105] The phase of the signals can be advanced, for example, according to the following formulas which are well known in the field of signal processing:

[0106] -the Taylor series formula, especially when the signal frequency bandwidth is low.

[0107] - the Padé formula order (m, n) with limiting thresholds on the derivative of the first and second signals, m being the order of the numerator and n the order of the denominator

[0108] According to a Fourier series or FFT for "Fast Fourier Transform" in English. A signal is decomposed into a periodic Fourier series.

[0109] We then know for a signal the compositions in frequencies and amplitudes and phasing with a window applied to the spectrum in order to filter the low frequency components and select a band or the frequencies which interest us to advance the signals.

[0110] We are interested in a frequency band and / or certain specific frequencies while eliminating the rest.

[0111] It is then possible to modify the amplitudes of different spectra in this frequency band and eliminate certain signals.

[0112] According to the invention, the correction of the delay comes from the expression of the delay according to the 2nd order Padé formula of the type (2,2):

[0113] [Math.7] 1 1 . ——- = expfe. sj =----~----- ï _ ± A J__±_ f * -12-

[0114] X' is the modified signal 12' of the delay correction,

[0115] X is the input reference signal 12.

[0116] The Padé formula is preferred to a Taylor expansion to broaden the frequency band. However, since the poles of the Padé formula (2,2) (the roots of the denominator function) have positive values, this causes the open-loop system to become unstable.

[0117] According to one embodiment of the invention, coefficients are introduced into said formula, in particular in the denominator, which are k3 and k4 in order to push the frequency of the poles towards much higher values, i.e.:

[0118] [Math. 8] X*(î) i + -y . s) 4- (¾. s)-; k3 ,■ :vT

[0119] s is a Laplace variable. To do a frequency analysis, we replace s by ico where co is the impulse equal to 2. pi times the frequency, i imaginary, or:

[0120] [Math.9] s = ioavee -â? =

[0121] ki, k2, k3 and k4 are tuning parameters or optimized by the least squares method. But k3 and k4 are low values ​​which vary with an order between 106 and 102 for a sampling step of 1ms.

[0122] According to another embodiment of the invention, threshold values ​​are introduced on the first and second order derivative when solving the discretized Padé equation (2,2) without coefficients ki to k4 to make the digital system stable.

[0123] This modification makes it possible on the one hand to broaden the frequency band of the signal compared to the development of the Taylor series, but also to make the algorithm more robust, in particular in the face of noise which is present in a complex system, such as a large driving simulator.

[0124] However, said modification results in the modification of the amplitude for high frequency signals. Corrections are then necessary depending on frequencies in order to widen the frequency band and filter out unwanted high frequency components,

[0125] This advance is added into the reference signal 12 in the first step 20.

[0126] During the first step, the additional sub-frame 19 comprises a fictitious sequence 19b to take into account the periodicity of the frame in order to improve the frequency resolution of the following second step which concerns the frequency decomposition of said modified reference signal.

[0127] According to one embodiment, said fictitious sequence 19b is a sequence 17 of zero values.

[0128] Preferably, the additional sub-frame is formed successively from a predictive sequence of prediction signals obtained by a transform or convolution by the Padé formula of order 2,2 of the reference signal, followed by a fictitious sequence of zero values ​​added at the start of the reference signal 12.

[0129] According to another embodiment, said fictitious sequence may be a symmetrical sequence 18 with respect to the time axis of the reference signal 12.

[0130] Thus the additional sub-frame 19 is formed successively from a predictive sequence of prediction signals 16 obtained by a transform by the Padé formula of order 2.2 of the reference signal, followed by a fictitious sequence symmetrical with respect to the time axis of the reference signal, added at the start of the reference signal.

[0131] According to another embodiment of the invention, the set of sub-frames of the preserved signal 12 can already be phase-corrected by the predictive formula. In this case, frame 19 is perfectly symmetrical.

[0132] We thus obtain a modified reference signal 12' with advance attributes as represented in [Fig.4].

[0133] According to the invention, it is then appropriate to reduce the modified reference signal 12' to a frequency band and / or certain specific frequencies by eliminating the remainder.

[0134] The method 10 comprises a second step 30 following the first step 20 of conversion into spectrum and phase of the modified reference signal 12'.

[0135] The second step 30 comprises a Fourier conversion with the FFT method for “Fast Fourier Transform” known in the field of signal processing.

[0136] The modified reference signal 12' is then converted into a frequency spectrum, i.e. into amplitudes and phases of the spectrum quickly.

[0137] The method comprises a third step 40 following the second step of filtering said spectrum.

[0138] To correct the modules that are amplified, filtering can be applied using one of the following methods:

[0139] -Introduce a low-pass filter or LP for "low-pass" in English in order to erase the high frequencies with a module which makes it possible to dampen the high module value of the corrected reference signal.

[0140] - Multiply the spectra with a coefficient less than 1 according to the identified frequencies.

[0141] To advance the phases of the signal by eliminating the low-frequency signal, the following two methods can be used:

[0142] -Introduce a high-pass filter allowing high-frequency components to pass in order to to erase the low frequencies.

[0143] -Multiply the spectra with a coefficient less than 1 according to the identified frequencies.

[0144] The first method is simple to implement, but it is less efficient than the second method in terms of phasing and the quality of the restored signals, because the phase advancement obtained by the first method is highly non-linear. The second method allows for better phase linearity to be maintained. On the other hand, a real-time FFT calculation with fine resolution requires greater effort.

[0145] Any other type of digital filter can be defined in the FFT Fourier conversion in order to dampen the influences of the high frequency signal.

[0146] In the third step, it is appropriate to exclude high frequencies, for example those above 5 Hz, and to reduce the influence of low frequencies.

[0147] Filtering can be performed with a weighting window 41 or weighting profile on the frequency spectra shown in [Fig.5]. The weighting window allows to reduce the amplitudes of the low frequency signals, on the other hand, the high frequency phenomena are well conversed. This non-linear scaling of the acceleration signal allows to adapt the capacity of the simulator to the movements to be reproduced by favoring a frequency band that interests us. This practice also allows to advance the signals because the low frequency components of the signal are reduced. It would be possible to correct the delay of the simulator by filtering the low frequencies. But the quality of the phasing is not good because of the phase distortion. This is one of the reasons why the predictive technique was also introduced in order to generate a reference signal with much less phase distortion.

[0148] An amplitude modulation profile can be defined as a function of frequencies. Said filter window can be designed either by a real filter using the spectra of the filter's impulse response, or by using the vestibular model which is a human-specific filter for motion detection, or by choosing the weighting coefficients manually.

[0149] The window of the modulation filter 41 takes the form according to [Fig.5].

[0150] It can be seen that low frequencies have a lower weighting factor and high frequencies or those above 4Hz are suppressed.

[0151] The phase advance can also be improved by modifying the weighting coefficients, particularly of the low frequencies.

[0152] With the predefined points of 16, one can also perform phase advancement with the convolution of inverse delay function directly in FFT.

[0153] Any other type of digital filter can be defined in order to dampen the influences of the high frequency signal.

[0154] The method comprises a fourth step following the third step of recomposing the reference signal whose spectrum is modified in the third step.

[0155] The signal is simply recomposed, for example, by taking the inverse FFT decomposition function. The sum of the Fourier series with modified amplitudes can then present a phase change.

[0156] Advantageously, the method makes it possible to easily modify the desired delay correction by simple parameterization.

[0157] The progress of the effective time is checked. The profiles of the initial or vehicle and recomposed reference signals are substantially identical and we seek to recalibrate the phase, for example by the intercorrelation function.

[0158] The filtering window can be modified, in particular by reducing the weighting factors of the low frequencies, or by improving the predictive sequence.

[0159] The objective is achieved: the method makes it possible to correct the delays due to the simulator in real time by adding an advance.

[0160] The predictive method is based on the Padé order (2,2) formula which has been discretized and thresholds of the velocity (first order derivative of the signal) and the acceleration (second order derivative) have been introduced in order to ensure that the predictive signals do not diverge or that the correction system is stable, because the Padé (2,2) formula is unstable for the system. Thus, compared to existing ones, the algorithm of our invention is more robust and the frequency band of the signal to be processed is greater.

[0161] Advantageously, the method performs a correction of the delay in real time, the parameters involved, such as the bandwidth, the gain coefficient for the low frequency signal are explicit, which facilitates the tuning task.

[0162] According to the invention, two combinatorial techniques are in fact proposed for compensating for the simulator delay.

[0163] A first technique is based on the predictive algorithm

[0164] A second technique is based on the reduction of the low frequency components of the reference signal.

[0165] These two techniques can be used separately or together. When it is necessary to reduce the low-frequency components due to the limitation of the simulator to reproduce the sustained acceleration, technique 1 is used. If it is necessary to reproduce the motion at scale 1, technique 2 is more suitable.

Claims

Claims

1. Method for correcting (10) the delay of a car driving simulator (50) for a motion restitution of a dynamic vehicle model (15) driven by a driver who is in said simulator (50) comprising a simulation platform capable of moving in a substantially planar workspace (53) according to an optimized and controlled trajectory following a route of the simulated vehicle in a virtual environment, said commands being optimized by a real-time algorithm (11) for the motion restitution, said platform is controlled by a central unit running said method which receives an input reference signal (12) and is capable of sending said modified signal (12') to the command algorithm (11) capable of generating commands to be passed to the animation means (57) of the simulation platform. characterized in that it comprises the following consecutive steps: - a first step of modification (20) of the frame of the reference signal (12) by adding predictive signals (16) to a part of said frame formed of the most recent last points of the reference signal, - a second step following the first step of identification of the modified reference signal (30) (12') in frequency distribution, - a third step following the second step of filtering the frequencies (40) of the modified reference signal (12') comprising a modulation profile (41) of the frequencies, - a fourth step following the third step of recomposition (50) of the input reference signal.

2. Method (10) according to claim 1, characterized in that during the first step (20), the frame of the reference signal (12) taken into account is reduced to the last points of said frame representing a maximum time sequence of 5 seconds.

3. Method (10) according to claim 2, characterized in that the first step (20) comprises an addition of an additional sub-frame (19) comprising the predictive signals (16) and arranged at the start of the reduced frame of the reference signal (12).

4. Method (10) according to claim 3, characterized in that the additional subframe comprises a sequence of predictive signals resulting from a transformation of the reference signal (12).

5. Method (10) according to claim 4, characterized in that the frame of predictive signals is added at the beginning of the additional subframe (19).

6. Method (10) according to any one of claims 3 to 5, characterized in that the additional sub-frame (19) comprises a sequence (17) of zero values.

7. Method (10) according to any one of claims 2 to 6, characterized in that the first step (20) comprises a generation of a frame which is composed of the reduced reference frame and an additional symmetrical part (18) of said reduced frame arranged at the start of the reduced reference frame (12).

8. Method according to any one of claims 2 to 7, characterized in that the third step comprises a modulation of the amplitudes of the frequency spectra.

9. Method according to any one of claims 2 to 8, characterized in that the third step comprises a modulation of the amplitudes of the combination of frequencies according to a modulation spectrum according to the frequency.

10. Method according to any one of claims 2 to 9, characterized in that it is provided to modify the correction of the delay by a parameterization based on an intercorrelation function.

11. Method according to claim 10, characterized in that the method comprises a phase of resetting the recomposed signal and the reference signal after the fourth step 50.

12. Method according to claim 10 or 11 in combination with claim 4, characterized in that the parameterization of the delay correction comprises a modification of the sequence of the predictive signals (16).

13. Method according to any one of claims 10 to 12 in combination with claims 8 and 9, characterized in that the parameterization of the delay correction comprises a modification of the modulation of the amplitudes of the frequency components of the modified reference signal (12').