METHOD FOR AUTOMATIC ADJUSTMENT OF AT LEAST ONE PERSONALIZATION PARAMETER OF A MOTOR VEHICLE

Description

[0001] The invention relates to a method for automatically adapting at least one parameter for personalizing the sound and light environment, posture, comfort and performance of a motor vehicle.

[0002] In the automotive field, a general issue is to seek to improve the personalization of the motor vehicle for the driver, and to offer an automation of this personalization.

[0003] This type of personalization can range from driver installation at the driving position when the driver is welcomed into their car (for example, electric seat adjustment or recognition of the user profile), to adapting the settings of the ambient systems (HMI and posture) and automotive dynamics (engine, chassis) while driving.

[0004] Several driving modes are frequently offered to a driver, such as a Normal mode, a Sport mode, or a modeEco (which minimizes the vehicle's energy consumption). Each mode applies different driving parameters to the vehicle, including engine performance parameters and gear ratios, in order, for example, to adapt to the driver's expectations of dynamic driving.

[0005] Drivers can change driving modes by typically pressing a button or other human-machine interface control, for example to select sport mode to increase engine power for sporty driving or to select eco mode for more fuel-efficient driving.

[0006] However, it is observed that users make relatively little use of these driving modes because their effect is not always fully understood. This results in underutilization of the potential benefits of personalizing vehicle settings according to their driving preferences.

[0007] Self-adaptive systems have been developed to estimate vehicle performance settings based on driver behavior data.

[0008] Notable prior art includes patent DE102018208431A1, which discloses a method for predicting the functional settings of a motor vehicle component based on usage data recorded during at least part of a journey. A machine learning process is used by a predictive device to automatically perform the functional adjustment. Other prior art is disclosed in EP 3 348 964 A1, DE 10 2011 117025 A1, and DE 10 2020 107536 A1.

[0009] The main limitation of using recent behavioral data to select vehicle parameters is that by considering only the past, a change in driving mode usually comes too late to be appropriate for a driver given the road or traffic situation.

[0010] The resulting time and geographical gap between estimated and actual driver needs leads to decreased driver acceptance, meaning that the potential of driving modes will remain underutilized.

[0011] This situation is exacerbated by the fact that current systems are unable to learn from the driver's ongoing behavior in order to improve automatic adaptations to specific needs in the use of driving modes.

[0012] Therefore, there is a need for a solution to improve the automatic adaptation of at least one parameter for customizing a motor vehicle.

[0013] To this end, we propose a method for automatically adapting at least one personalization parameter of a motor vehicle according to claim 1.

[0014] Thus, it is possible to automatically adapt a vehicle personalization parameter by anticipating the user's needs through knowledge of dynamic parameters, based on mapping, or based on information transmitted from a plurality of other vehicles that have previously traveled the same route.

[0015] Also, we can further improve the personalization of the process by classifying the driver according to a type of profile, so as to acquire previous dynamic data corresponding to profiles similar to that driver in order to anticipate more precisely his future driving.

[0016] The future position of the vehicle can be understood as any future position, depending on the direction of travel of the vehicle and map data, with an anticipation distance that is a function of the speed of the motor vehicle.

[0017] In an implementation, anticipation is carried out for the next 30 seconds, which translates into an anticipation distance, for example, between 500m and 3000m, for example 2000m, depending on the current and expected speed of the vehicle.

[0018] Alternatively, the future position can be determined based on the speed of the motor vehicle. For example, one can consider the positions that can be reached within a range of 1 to 30 seconds.

[0019] In particular, said at least one customization parameter includes a value for the available power of the motor vehicle.

[0020] In particular, said at least one dynamic parameter of a plurality of other motor vehicles includes at least one parameter from among: longitudinal acceleration, lateral acceleration, vertical acceleration, velocity, or a combination of at least two of these preceding parameters.

[0021] In particular, said at least one dynamic parameter of a plurality of other motor vehicles includes at least one parameter among an ambient lighting setting, an ambient sound setting, or a driver posture setting,

[0022] For example, said at least one dynamic parameter of a plurality of other motor vehicles includes at least one parameter among a cockpit posture, a sound broadcast in the passenger compartment, a color of lighting of the passenger compartment or of the exterior of the vehicle, a suspension setting.

[0023] Although the main implementation of the invention is based on the longitudinal dynamics of the vehicle, it is also possible that the method can be implemented for the lateral and vertical dynamics of the motor vehicle using performance parameters representative of the respective dimensions.

[0024] Advantageously, and without limitation, said at least one dynamic parameter is the product of velocity and acceleration over a predetermined distance from the current position of the motor vehicle. Thus, a relatively relevant parameter can be obtained that is representative of a power demand.

[0025] Specifically, one can acquire either a dynamic parameter for each vehicle and for each position, which increases the amount of data transmitted, or acquire an average value or a distribution of pre-calculated dynamic parameters to reduce the amount of data received. As we will see later in the description, these parameters can also be grouped, for example, by driving style (sporty, normal, "eco," for example), in order to obtain values ​​differentiated according to the driving style of the driver for whom we wish to anticipate driving parameters.

[0026] In other words, if the dynamic parameter stored in the database for each map position is a suitable power value (for example, a value of the product of acceleration and speed), for a plurality of motor vehicles, or in the form of an average value or distribution, then an average value of the power data generally requested by drivers is stored at each point on the road map. Consequently, a driver's needs can be anticipated over a future period, for example, between 1 and 30 seconds for the next 10 seconds, by scanning future positions on the map based on their geolocation.

[0027] Advantageously, and without limitation, the method further includes a step of storing the dynamic parameters of the motor vehicle at each positioning location on the road map, so that during a subsequent pass, said acquisition step also acquires these previous dynamic parameters of the motor vehicle at the same location on the road map. Thus, the performance parameters of the motor vehicle can be anticipated based on the previous dynamic data of the same driver.

[0028] Advantageously, and without limitation, the adaptation stage includes a feedback loop, for example, but not limited to, a negative feedback loop, during which the following is implemented: a step of receiving action data from the driver of the motor vehicle; a step of comparing the actions with said at least one suitable personalization parameter calculated in the calculation step, the command step being a function of said comparison carried out during the comparison step.

[0029] This allows the process to improve the accuracy of anticipation by taking into account the driver's intentions, and not to go against their request or safety, but only to support their driving.

[0030] The invention also relates to a device, such as an embedded computer, for the automatic adaptation of at least one personalization parameter of a motor vehicle according to claim 8.

[0031] All the means implemented by the device, such as positioning, acquisition, calculation or control means, can be an embedded computer, a processor, a processor core, or any other digital computing device.

[0032] Furthermore, this device includes, in particular, RAM for storing temporary data.

[0033] This device may further include a storage memory to store all or part of said road mapping and / or all or part of the database of at least one dynamic parameter of a plurality of other motor vehicles.

[0034] The device may also include telecommunication means, such as cellular, GSM, satellite, or any other suitable means to communicate with a remote server, for example to acquire all or part of said road mapping and / or all or part of the database of at least one dynamic parameter of a plurality of other motor vehicles, and to also transmit the dynamic data of the motor vehicle as well as the relative positive acceleration data possibly calculated.

[0035] The invention also relates to a motor vehicle comprising a device as described above.

[0036] Other features and advantages of the invention will become apparent from the following description of a particular embodiment of the invention, given by way of example but not limitation, with reference to the attached drawings in which: [ Fig. 1 ] is a schematic view of a first embodiment of the invention; [ Fig. 2 ] is a schematic view of a second embodiment of the invention; and [ Fig. 3 ] is a functional view of an example embodiment of the process according to the invention.

[0037] A motor vehicle includes an on-board computer adapted to modify at least one personalization parameter of a motor vehicle, such as engine power settings.

[0038] This embedded computer, which is considered in the context of the invention to be a particular type of computer, implements a method 1 for automatically adapting the personalization parameter.

[0039] This method 1 according to the first embodiment of the invention implements a so-called proactive loop.

[0040] The term loop refers to a set of steps in the process.

[0041] The proactive loop predicts the desired vehicle settings and adapts the vehicle personalization parameter before the driver is expected to become aware of this need.

[0042] To this end the process first implements a geolocation step (10) of the motor vehicle.

[0043] The motor vehicle includes, in particular, a GPS, or other geolocation device, enabling the acquisition of the vehicle's coordinates on the globe. This geolocation device is adapted to communicate geolocation data to the computer implementing the process, either on demand or by transmitting it to the computer at regular intervals.

[0044] Next, we proceed to a step of positioning the motor vehicle on a road map according to said geolocation.

[0045] For this purpose, a road map is stored in the vehicle's memory, which can be read by the computer.

[0046] In this embodiment, this mapping is stored in mass memory, but it can alternatively be acquired as needed via a telecommunications network on a remote server.

[0047] Thus, the geolocation position is correlated with the road map to position the vehicle on a road. This positioning step can also determine the vehicle's direction on the road when several successive geolocation positions of the vehicle are available.

[0048] Once the vehicle is positioned on the road map, we then proceed to an acquisition step 12 in a database of at least one dynamic parameter of a plurality of other motor vehicles in the same position of the motor vehicle on the road map.

[0049] The database thus stores a plurality of dynamic parameters of vehicles that have taken the same "route" on the road map, whether it be other vehicles or the motor vehicle implementing the process during a previous passage.

[0050] This step is carried out in this embodiment by acquiring data from a database stored in the memory of the motor vehicle.

[0051] But according to an alternative embodiment this database can be accessed remotely, for example via a remote server connected by a telecommunications network.

[0052] In this embodiment, the dynamic parameters are calculated as the product of speed and acceleration, which is a relevant indicator of the power demand made by the driver.

[0053] Dynamic parameters may, however, include speed data, longitudinal acceleration data of other motor vehicles, lighting or sound environments, driving postures, suspension settings, or of the motor vehicle during previous passes.

[0054] But they can also include, alone or in combination, data on instantaneous speed, lateral acceleration, cockpit posture and chassis agility, particularly in the context of four-wheel steering vehicles.

[0055] According to an alternative implementation, the dynamic parameters are acquired in the form of an average value from all other motor vehicles, which in particular reduces the amount of data stored and / or processed by the process.

[0056] Starting from this data, we then proceed according to these dynamic parameters acquired in the calculation step 13 of the personalization parameters according to the said acquired values ​​of the said at least one acquired dynamic parameter values.

[0057] For the purposes of this invention, personalization parameters are parameters that have an effect on the driving performance of the motor vehicle or on comfort (particularly with regard to lighting or sound environments, driving postures, suspension settings)

[0058] By analogy, personalization parameters are the parameters modified when switching from eco mode to sport mode or sport mode of a motor vehicle, each mode modifying a set of personalization parameters of the motor vehicle to modify, for example, its dynamics, its available power, and its comfort.

[0059] In particular, process 1 can adapt the same parameters as those modified when changing a driving mode (economic, usually abbreviated "eco", sport, normal) of a motor vehicle by driver selection.

[0060] Then we proceed to command step 14 of the application of the adapted personalization parameter calculated in step 13 to the motor vehicle.

[0061] This control step 14 thus makes it possible to anticipate the behavior of the driver of the motor vehicle in relation to the knowledge of the behavior of other drivers in the same place on the road map and to modify the driving parameters of the vehicle accordingly.

[0062] Thus, process 1 implements a proactive "loop", in other words a set of proactive action steps, to predict the desired driving settings of the vehicle and adapt these vehicle settings just before the driver is expected to feel the need, based on an anticipatory process that uses data from a database associating driver personalization parameters with geographical locations.

[0063] A second embodiment of the invention is an evolution of this first embodiment, but aims to take into account the behavior of the driver in addition.

[0064] Indeed, in the first embodiment, the adjustment of driving personalization parameters is done solely based on knowledge of the behavior of other drivers, and possibly of the driver himself during past journeys, stored in a database.

[0065] However, this process 1 can be enhanced by also taking into account the behavior of the car driver during driving.

[0066] Thus, process 1' of the second embodiment implements the steps of the first process 1 while also integrating feedback control steps, including negative feedback.

[0067] Process 1', in reference to the figure 2 , receives, during an acquisition step, data on the driver's actions on the motor vehicle.

[0068] These action data include, for example, a request for acceleration or braking, lateral acceleration, a gear change action, a change in driving posture, a change in light and sound environment, an action on the steering of the motor vehicle, or any other action modifying the behavior of the motor vehicle.

[0069] In this way, after implementing the geolocation 10, positioning 11, and acquisition 12 steps, we proceed to the proactive adaptation calculation step 13 as carried out during process 1, but in this embodiment the adaptation is not immediately implemented by the control step 14.

[0070] Indeed, during a reception step 15, we receive the driver's actions data for the motor vehicle, then in the comparison step 16, we compare these received actions with the adaptation calculated in step 13.

[0071] Depending on the result of the comparison, a command step 14 of the personalization parameter is then implemented based on the performance parameter values ​​acquired in step 12 and the values ​​received in step 15 of the driver's actions.

[0072] The objective of this comparison is to ensure that proactive adaptation does not conflict with the driver's intentions or safety.

[0073] Steps 14 to 16 thus form a negative feedback loop to eliminate misalignments between the proactive loop and the vehicle setting desired by the driver.

[0074] Misalignments, in other words incompatibilities between the proactive action and the action performed by the driver, can here be eliminated immediately, either implicitly by canceling the proactive prediction if the current behavior of the driver deviates sufficiently from the currently estimated inputs, or explicitly by the driver by pressing a control button (for example, a manual selection of the mode, change of posture, change of light or sound environment).

[0075] Thus, long-term misalignments can be eliminated and predictions improved; the process also implements an update step 17 of a driver behavior and personalization parameter database, which can be either the same database as for the dynamic parameter data of other drivers, or another database, for example stored locally in the motor vehicle, but can also be stored remotely, to record the anticipated driving behavior in the database with the individual driving data of the recorded driver, specific to the location, which allows a personalized and more specific adaptation of the driving personalization parameter, and learns the individual's preferred proactive driving personalization parameters.

[0076] Another feature of the method according to the invention that can be implemented starting from the second embodiment of the invention or its variants, consists of calculating a value representative of the type of driving of the motor vehicle user in order to provide him with an engine power that also takes into account his usual type of driving.

[0077] For this purpose, an indicator called relative positive acceleration, known in English by the abbreviation of Relative Positive Acceleration, RPA. This value, known to those skilled in the art, corresponds to the integral over a time interval of the vehicle's speed multiplied by its positive acceleration, divided by the total distance traveled in the interval. It is thus an indicator of the driver's power demand over time.

[0078] Thus, in a particular embodiment of the invention, the driver's RPA is calculated, for example by taking into account the recorded dynamic values ​​as previously described.

[0079] This RPA value is then used during acquisition step 12, so as to discriminate the acquired dynamic parameters, to acquire only the parameters associated with recordings of drivers having a similar RPA.

[0080] For this purpose, ranges of RPA values ​​can be predetermined, for example according to usual criteria: sporty driving (e.g. high RPA), “normal” driving (e.g. medium RPA) or “eco” driving (e.g. low RPA).

[0081] These types of driving classified by RPA (as a driver profile) can also be differentiated according to the types of roads (Highway, national roads, city).

[0082] RPA can also be calculated over a short period, for example to estimate the driver's driving reactivity (nervous, calm) as well as over a long period, for example to estimate the driving mindset (calm, hurried).

[0083] In this way, the calculation of proactive driving parameters can be carried out while also taking into account the driver's usual driving style. In other words, the calculated RPA value makes it possible to determine a driving profile for the user, corresponding to a range of RPA values.

[0084] This RPA value can be updated regularly by the process, for example during database update 17.

[0085] A feedback loop, represented on block 19 of the figure 3 can be implemented to adapt the driver-associated RPA proactively predicted and applied as a personalization tool to engine power to the current RPA measured while driving.

[0086] RPA values ​​can be associated with geographical coordinates for the user, times of day, or any other relevant criteria.

[0087] Thus, a machine learning module can be implemented that is capable of determining the expected RPA of the driver, based on their geographical location and / or other relevant criteria.

[0088] The process according to any of the preceding embodiments may also take into account external parameters, and may discriminate the values ​​acquired in step 12 according to these parameters, such as the time of day, weather information, traffic information, or any other external information that may influence driving dynamics and the personalization of the ambiance, posture, and comfort of the motor vehicle.

[0089] With reference to the figure 3 , a preferred embodiment of the invention is described according to a functional block approach.

[0090] There figure 3 is formed by five interconnected blocks: {1a, 1b}, 2, 3, {4a, 4b} and 5, which are described below. Although each block has its own contribution, the overall performance of the process according to the invention relies on the interconnection of the blocks.

[0091] On the figure 3 Each block is also associated with the main steps of the process to improve understanding.

[0092] The first block {1a, 1b} includes a module called the predictive module for adjusting vehicle driving personalization parameters.

[0093] This predictive module, which corresponds to steps 10-13 of the process, proactively proposes an adjustment of the vehicle driving personalization parameters using a decision prediction based on the driver's anticipated behavior according to location, i.e. the driver's expected behavior in the future given the location, and the driver's current behavior.

[0094] As an example, during driving, the predictive module searches a few seconds, for example within a prediction window in a range of 1 second to 20 seconds, for example 10 seconds ahead along the route in the database with location-specific driving behavior.

[0095] To do this, the predictive module examines the driver's route over a prediction time window, for example, through their navigation system or any other suitable means to estimate the most likely route given their general behavior, then selects a number of GPS locations and extracts from the database the driving behavior corresponding to these locations.

[0096] The behavior in the prediction window is selected and used to determine the customization setting of the engine and / or gear ratios.

[0097] Block 1a corresponds to the anticipated behavior of the driver depending on the location.

[0098] This location-specific anticipated driver behavior consists of behavioral variables and driver personalization choices, such as longitudinal / lateral acceleration, speed, accelerator and steering input, ambient lighting and sound preferences, and posture choices, all linked to the geographic location. Anticipated driver behavior can be based on historical data (a priori measurements of driver behavior, similar to heat maps or traffic monitoring).

[0099] Anticipated driver behavior can also be predicted using numerical driver behavior models and road parameters, e.g., curvature, type, expected traffic information.

[0100] The driver's anticipated behavior implicitly takes context into account, for example, curves, stop signs, speed bumps, and the surrounding environment (the presence of a forest, the sea, a school, or a place of worship), unlike the explicit inference of context by the vehicle's sensors. Indeed, a stop sign, for example, is implicitly recognized by the deceleration and subsequent acceleration of drivers in its vicinity.

[0101] The driver's anticipated behavior can thus be personalized once the driver starts driving with the system; the system records and updates the driver's anticipated behavior based on location.

[0102] In this implementation example, the predictions are based on the anticipated behavior of the driver depending on the location, operationalized by the speed * the positive acceleration averaged over a distance, obtained by geolocation, here GPS, along the driver's route; which constitutes a criterion equivalent to the energy demand.

[0103] We assume that we have access to a database on location-specific driving behavior: Either the driver has already made the journey before and their data has been recorded and matched with GPS locations, or we have a database with the behavior and choice of personalization parameters of other drivers.

[0104] If no data is available for the desired locations, driver behavior models could be used to make predictions.

[0105] In addition to data relating to driver behavior, including the use of the accelerator and brake pedals, gear changes and the use of the steering wheel, and dynamic vehicle parameters, such as longitudinal and lateral accelerations or vehicle speed, we also record self-selected parameters, such as the pressing of a button, driving posture choices, lighting and sound ambiance choices, and distance travelled.

[0106] Data storage takes place in two phases.

[0107] First, the driver's behavior over time is stored in a 100 Hz buffer. Then, the data is resampled at a GPS location and written to a database.

[0108] The behavior of the newly registered driver is weighted by the average of the data already present in the database (with less weight for new data).

[0109] In other words, the database consists of rows of GPS positions and columns of driving behavior variables (speed, acceleration, distance traveled, and accelerator pedal position). If a new GPS location is found that was not already in the database, it will be added.

[0110] The data memory only stores the speed and acceleration values ​​given by the location; therefore, a more precise adjustment of the vehicle's settings will not affect the data memory.

[0111] Indeed, in a preferred mode we have speed and acceleration separately to calculate the anticipated power on board.

[0112] Furthermore, acceleration and velocity are more readily available from multiple sources.

[0113] Block 1b corresponds to the vehicle setting predictor, also known as driving parameters.

[0114] The proactive adaptation process integrates the driver's anticipated and current behavior into the proposed vehicle settings using decision-making logic.

[0115] Finally, in addition to the implicit contextual information arising from the anticipated driving behavior (block 1a), the comparison between the anticipated behavior and the current behavior implicitly takes into account the dynamic elements of the road (e.g., traffic lights, traffic).

[0116] For example, it is likely that the proactive adaptation process predicts a sportier engine setting near a traffic light, because on average drivers decelerate and then accelerate sharply (when the traffic light is red).

[0117] However, if the light is green and there is a discrepancy between the predicted behavior (red light; deceleration followed by acceleration) and the driver's actual behavior (green light; constant speed), the vehicle's settings prioritize the driver's current behavior, meaning it remains in the vehicle's current settings. Therefore, there is no automatic switch to a sportier engine setting.

[0118] During driving, the process searches a few seconds in advance along the route in the database with location-specific driving behavior.

[0119] To do this, the process examines the driver's route over the prediction window (e.g., via their navigation system or to assume the most likely route given their general behavior), selects a number of GPS locations and extracts from the database the behavior and driving personalization choices corresponding to these locations.

[0120] The behavior in the prediction window is selected and used to determine engine tuning and customization choices.

[0121] If the value of the dynamic parameter, for example the product speed*acceleration, in the prediction window exceeds a certain threshold, i.e. when a high power demand is expected, the vehicle parameters switch for the engine to the high engine power setting or even for the color of the ambient lights, for example to bright and dynamic color.

[0122] In this example, we only use velocity*acceleration (also written acc_pos, i.e., acceleration for the given position) given the location as a prediction signal.

[0123] For longitudinal prediction, we examine the power / energy demand, which is force x velocity, or acceleration x velocity, to which the vehicle's mass could also be added. The reasoning behind this is that we are looking for situations where drivers require high engine power. For these situations, we change the engine parameters for a more "aggressive" setup, i.e., a more responsive and readily available throttle, higher RPM shifting, and so on.

[0124] Finally, the vehicle placement prediction system has several adjustable customization parameters. In summary, the policy parameters are as follows: The decision threshold for adapting the vehicle's personalization settings is a particularly important adjustable parameter, as explained below. The prediction window size refers to the length of time the system projects itself into the future to make appropriate personalization predictions. This involves analyzing and using the amount of anticipated driver behavior data to decide if and when to change personalization settings, such as engine power responsiveness. A shorter prediction window results in shorter, more frequent changes to personalization settings. The duration of the personalization change is also a factor.Once a customization setting is changed, for example, to a high engine power setting, this duration determines how long the vehicle's customization setting must remain in that setting (e.g., longer in a more "aggressive" engine power mode) before automatically reverting to the setting prior to the change (e.g., a low engine power setting). The minimum switching time is the minimum amount of time that must elapse between two consecutive customization setting changes. If two scheduled switches fall within this timeframe, the algorithm maintains the current customization setting. This check prevents the system from switching too frequently.

[0125] Regarding the setting of the decision threshold for adapting the vehicle's personalization settings, and as an example, a decision threshold has been implemented as rule-based logic for switching from eco mode to sport mode (but this can be generalized to transitions between other driving modes). Thus, in this particular example, the setting consists of changing the personalization parameter between the preset driving modes (eco, sport, normal).

[0126] To predict, for example, the anticipated power demand, in other words the power demand in the immediate future of the route ahead of the vehicle, the process is based on the driver's behavior specific to the location being observed.

[0127] For the purposes of this invention, the term "driver" refers to the current driver of the motor vehicle, and the latter can be previously identified in a dedicated step by means of a known method.

[0128] This allows, in particular, for the vehicle to be personalized to the current driver, especially in the case of a vehicle shared by several different drivers.

[0129] Here, the prediction parameter used for the anticipated power call is the dynamic parameter within the prediction window, integrated over the window and normalized by the distance traveled within the window. This measure is also called relative positive acceleration. RPA. There RPA is also correlated with fuel consumption and driving behavior.

[0130] However, other prediction parameters based on GPS location can be used, for example, the power delivered by the vehicle directly, acceleration, jolts or combinations of these prediction parameters, sound and light environments, postures, suspension comfort choices.

[0131] The decision logic for adapting vehicle personalization settings is threshold-based, as in the example of engine power customization. When the decision parameter reaches and exceeds a threshold, it indicates a period of high power demand, and the vehicle settings are adjusted accordingly. Monitoring this threshold is also used to decide whether to change driving position, increase brightness, or enhance the ambient sound.

[0132] The magnitude of the threshold has a direct influence on the duration and frequency of changes in driving personalization parameters, which we will illustrate with two representative examples (a high threshold and a low threshold).

[0133] Firstly, a high threshold will result in fewer and shorter adjustments to the vehicle's settings, for example, for engine customization, at high power.

[0134] A low threshold will result in more frequent and longer periods of switching between driving personalization parameters, for example, the engine with a higher power personalization.

[0135] The threshold amplitude also has a direct impact on the frequency and duration of vehicle setting changes.

[0136] Given the differences in driving preferences between drivers, the threshold can and should be adapted to each driver. It's also possible to consider that several thresholds will be offered depending on the types of personalization desired: thresholds for ambiance, thresholds for posture, etc.

[0137] Any threshold can be adjusted based on explicit driver interactions (including manual increase or decrease of the threshold by the driver, or selection from a list of predefined thresholds) or by implicit adjustment of the threshold based on driver behavior.

[0138] In the latter case, the thresholds can be adjusted according to driving behavior; if the driver's behavior resembles sporty driving, the threshold can be automatically lowered. If the behavior resembles that of a more conservative driver (few gear changes, for example), the thresholds can be raised.

[0139] Block 2 represents a module for immediate misalignment reduction using a reactive intention detector.

[0140] The reactive intent detector serves as input for arbitration of vehicle personalization parameter settings and helps eliminate immediate misalignments between the predicted proposed vehicle setting and the observed current driver intent.

[0141] The vehicle's reactive tuning is calculated using decision logic based on the driver's current behavior (including pedal use, gear changes and steering wheel use, changes in ambient sound and light, changes in posture, suspension settings) and the vehicle's condition (i.e., longitudinal and lateral accelerations or vehicle speed).

[0142] Block 3, on the other hand, represents the arbitration of vehicle parameters.

[0143] Vehicle configuration arbitration consists of a predictive calculation that decides the final vehicle customization configuration by taking into account the customization configuration proposed by the proactive adaptation process and the reactive vehicle configuration.

[0144] Vehicle setup arbitration compares, as in step 16 of the process, the driver's predicted intention with the driver's observed current intention and selects the vehicle setting that best supports the driver's behavior.

[0145] Finally, if necessary, the driver's request for manual adjustment will always take priority, allowing the driver to override the system if they wish.

[0146] Block 4 represents the minimization of long-term misalignments of personalization.

[0147] While the reactive intent detector aims to eliminate immediate misalignments between the driver's intention and the predicted parameters proactively, the personalization approach aims to learn from these misalignments of the personalization parameters in order to improve future predictions and prevent the same misalignment from recurring. Furthermore, personalization correction is essential to making predictions predictable for an individual driver. Personalization correction involves two elements: According to block 4a, the driver's anticipated personalization choices are learned. The driver's anticipated personalization choices are then updated with the recorded current behavior. This learning process is typically long-term.The more a route is traveled, the more relevant the driver's anticipated, location-specific personalization choice behavior becomes.

[0148] Block 4b represents the learning of personalized proactive prediction parameters, also known as driving personalization parameters.

[0149] Preferred proactive personalization settings are learned using observed driver behavior and an explicit driver request (such as a manual mode change using a button, a change in lighting ambiance, or a posture).

[0150] Decision logic can learn from misalignments between its own predictions of personalization choices and the observed ongoing behavior (of driving and / or personalization choices) of the driver by updating the parameters of the proactive prediction.

[0151] It is relatively important that the decision logic leads to predictable personalization choice decisions for the driver in order to avoid disaffection (i.e. the choice of new personalization settings that would replace the current settings); in particular, if the predictive adaptation does not adapt to a repeatable situation, the driver may not accept the current settings of the personalization system with which the vehicle is equipped.

[0152] When a proactive adaptation is finally ordered according to step 14, it is also possible to proceed as described in block 5 with a communication of the intention of the system whose setting has changed (for example by explaining the reason for the change in cockpit setting) and the estimated intention of the driver.

[0153] Decisions must be explainable and logical to the driver, hence the importance of adequate communication of the decision-making process to improve the driver's understanding of how the system works and, subsequently, their acceptance.

[0154] The system's intentions are communicated to the driver through sound, light, and the dashboard interface.

[0155] Communication consists, for example, of: Communication of the mode change. Communication of the adjustment intention calculated by the process. Communication of the driver's estimated intention by the process. Communication of observed misalignments and learned adjustments.

[0156] The method according to the invention thus allows for proactive adaptation of vehicle personalization parameters, such as the vehicle's longitudinal, lateral, and vertical behavior, as well as ambient parameters, posture, etc., to the individual and location-specific needs of the driver. The main advantages of this system are as follows: The method according to the invention describes a holistic process enabling the vehicle settings to be proactively adapted according to the driver's behavior.

[0157] Driver behavior is determined by the driving location, such as location-specific behavior, as well as by the driving context, such as traffic conditions. Therefore, signals are location-specific and may include driver behavior or vehicle condition, such as power output.

[0158] The vehicle's parameters are proactively modified, based on predictions, before the driver even feels the need. This ensures the vehicle is ready so the driver can immediately benefit from the adjusted settings when required.

[0159] Thus, once the route has been followed by the process as part of the search for proactive driving parameter settings, the location-specific anticipated driver behavior database is updated with the driver behavior of the recorded individual.

[0160] The behavior of the newly registered driver is weighted on average with the data already present in the database, in particular with less weight for new data.

[0161] This automatically results in more personalized predictions the next time the driver takes the same route.

[0162] The combination of proactive predictions, reactive intention detections, and learning allows us to have predictions accepted by the driver, because short- and long-term misalignments are eliminated by prioritizing the driver's current intention over the predicted intention and by learning from observed driver behavior.

[0163] The system eliminates immediate discrepancies between the driver's predicted intention and the driver's actual intention by comparing the vehicle's predicted parameters with the driver's actual intention; in other words, the process implements a reactive intention observer with the possibility for the driver to override it.

[0164] This allows the vehicle settings to be customized by learning the individual driver's behavior and preferences in order to minimize future and repetitive misalignments.

[0165] Appropriate communication also helps to inform the driver and eliminate confusion about the vehicle settings.

[0166] The method according to the invention not only switches from one predefined static driving mode to another, but also allows driving parameters to be adjusted according to the driver. Vehicle settings and combinations of vehicle settings (i.e., different settings for the vehicle's longitudinal and lateral dynamics based on individual preferences, ambient lighting and sound settings, posture, or road conditions) can be dynamically adapted during driving based on the driver's historical, current, and anticipated behavior. The vehicle settings can adapt to a continuous range of settings and combinations of vehicle settings.

[0167] No other advanced driver assistance sensors are needed to make reliable predictions. Because the predictions are based on location-specific driver behavior recorded a priori, our approach reduces reliance on behavioral or traffic models. The process is thus able to implicitly infer dynamic road situations, such as traffic and traffic lights, from anticipated and actual driving behavior.

[0168] The system does not necessarily rely on a predefined classification of driving style, road types, or situations to adapt vehicle parameters. It relies directly on the individual driver's behavior, allowing for a more personalized and timely adaptation of vehicle parameters.

[0169] Furthermore, the invention requires few adjustment parameters, and minimal adjustments are necessary. Key performance indicators of the proactive adaptive system, such as the frequency, duration, timing, and location (start / end position) of vehicle parameter adaptations, emerge automatically from the driver's personalized behavior based on location.

Claims

1. Method (1, 1') for automatically adapting at least one personalization parameter of a motor vehicle, such as a parameter for personalization of the engine, of the chassis, of the lighting and sound ambiance of the passenger compartment, of driver posture, said method comprising: - a step (10) of geolocating said motor vehicle; - a step (11) of positioning the motor vehicle on a road map on the basis of said geolocation; - a step (12) of acquiring, in a database, at least one dynamic parameter of a plurality of other motor vehicles in the same position and / or in future positions of the motor vehicle on the road map; - a step (13) of calculating said at least one personalization parameter adapted on the basis of said at least one acquired dynamic parameter; and - a step (14) of commanding the application of the personalization parameter adapted to the motor vehicle; characterized in that the method further comprises: - a step of calculating a value representative of the type of driving of the user of the motor vehicle, such as a relative positive acceleration value over a predetermined period; - the definition of a driving profile of the user based on said value representative of the type of driving of the user; and acquired in said database during the acquisition step (12) is at least one dynamic parameter of a plurality of other motor vehicles having the same value representative of the type of driving of the user of the motor vehicle, calculated for said motor vehicle and for the other vehicles, in the same position of the motor vehicle on the road map.

2. Method (1, 1') according to Claim 1, characterized in that said at least one personalization parameter comprises an available power value of the motor vehicle.

3. Method (1, 1') according to Claim 1 or 2, characterized in that said at least one dynamic parameter of a plurality of other motor vehicles comprises at least one parameter amongst: longitudinal acceleration, lateral acceleration, vertical acceleration, speed, or a combination of at least two of these parameters above.

4. Method (1, 1') according to any one of Claims 1 to 3, characterized in that said at least one dynamic parameter of a plurality of other motor vehicles comprises at least one parameter amongst a light ambiance setting, a sound ambiance setting or a driver posture setting.

5. Method (1, 1') according to Claim 4, wherein said at least one dynamic parameter comprises the product of the speed and the acceleration over a predetermined distance from the current position of the motor vehicle.

6. Method (1, 1') according to any one of Claims 1 to 5, characterized in that it further comprises: - a step of storing the dynamic parameters of the motor vehicle at each positioning location on the road map so that, during a subsequent passage, said acquisition step (12) also acquires these previous dynamic parameters of the motor vehicle, at the same location of the road map.

7. Method (1') according to any one of Claims 1 to 6, characterized in that the adaptation step comprises a negative feedback loop during which are implemented: - a step of receiving action data from the driver of the motor vehicle; - a step of comparing the actions with said at least one adapted personalization parameter calculated in the calculation step (13), the command step (14) being based on said comparison carried out during the comparison step (16).

8. Device for automatically adapting at least one personalization parameter of a motor vehicle, such as a parameter for personalization of the engine, of the chassis, of the lighting and sound ambiance of the passenger compartment, of driver posture, said device comprising: - means for geolocating said motor vehicle; - means for positioning the motor vehicle on a road map on the basis of said geolocation; - means for acquiring, in a database, at least one dynamic parameter of a plurality of other motor vehicles in the same position and / or in future positions of the motor vehicle on the road map; - means for calculating a value representative of the type of driving of the user of the motor vehicle, such as a relative positive acceleration value over a predetermined period; - means for defining a driving profile of the user based on said value representative of the type of driving of the user; and - means for calculating said at least one personalization parameter adapted on the basis of said at least one acquired dynamic parameter; and - means for commanding the application of the personalization parameter adapted to the motor vehicle; the acquisition means being designed to acquire, in said database, at least one dynamic parameter of a plurality of other motor vehicles having the same value representative of the type of driving of the user of the motor vehicle, calculated for said motor vehicle and for the other vehicles, in the same position of the motor vehicle on the road map.

9. Motor vehicle comprising a device according to Claim 8.