FR3057517A1 - DEVICE FOR PREVENTING DANGEROUS SITUATIONS FOR A CONDUCTOR OF A TRANSPORT VEHICLE AND ASSOCIATED METHOD - Google Patents

Abstract

The invention relates to a device for preventing (100) dangerous situations for a driver of a transport vehicle, comprising: - a memory (30) in which is stored a behavioral diagram of the driver during his driving, this behavioral diagram including data relating to at least one dangerous driving situation of the driver related to an insufficient driver's state of alertness; - a driver's driving situation determination module (20) adapted to determine a current driving situation of the driver, an estimation module (10) for a current state of vigilance of the driver, and a control unit (50) programmed to: compare the current driving situation of the driver with the dangerous driving situation stored in the diagram behavioral, - compare the current state of alertness of the driver to a state of vigilance threshold, and - order the issuance of a warning signal at destination of the driver according to said two comparisons.

Description

Holder (s): VALEO COMFORT AND DRIVING ASSISTANCE Simplified joint-stock company.

Extension request (s)

Agent (s): VALEO COMFORT AND DRIVING ASSISTANCE.

DEVICE FOR PREVENTING DANGEROUS SITUATIONS FOR A DRIVER OF A TRANSPORT VEHICLE AND ASSOCIATED METHOD.

FR 3 057 517 - A1 (5 /) The invention relates to a device for preventing (100) dangerous situations for a driver of a transport vehicle, comprising:

a memory (30) in which a behavioral diagram of the driver is stored during his driving, this behavioral diagram comprising data relating to at least one dangerous driving situation of the driver linked to an insufficient state of alertness of the driver,

a module for determining (20) the driving situation of the driver adapted to determine a current driving situation of the driver,

- an estimation module (10) of a current vigilance state of the driver, and

- a control unit (50) programmed for:

- compare the driver's current driving situation with the dangerous driving situation stored in the behavior diagram,

- compare the driver's current vigilance state to a threshold vigilance state, and

- order the emission of an alert signal to the driver based on said two comparisons.

Technical field to which the invention relates

The present invention relates to the prevention of drowsiness or distraction of drivers of transport vehicles.

It relates in particular to a device for preventing dangerous situations for a driver of a transport vehicle and to a corresponding method.

It finds a particularly advantageous application for drivers of motor vehicles.

TECHNOLOGICAL BACKGROUND

There are known monitoring devices for transport vehicle drivers adapted to alert drivers when their alertness is insufficient, for example when they are drowsy or distracted. Such an insufficient state of vigilance may indeed lead to a loss of control of the vehicle.

In general, such devices include a module for capturing images of the driver's head, and a control unit adapted to analyze these images to extract physiological information from the driver behind the wheel, such as the duration of the closing of his eyelids. or the direction of his gaze, and deduce his state of alertness.

There are also known monitoring devices which alert drivers when their alertness is deteriorating.

Document US20130021463 describes in particular such a monitoring device adapted to detect the alertness state of a driver preceding his drowsy state, and this by measuring the angular movement of his face and the duration of opening of his eyelids .

However, all of these devices only alert the driver when the insufficient vigilance state is already reached or very close to being reached. They only take into account this state of alertness to alert the driver.

Object of the invention

In order to remedy the aforementioned drawback of the state of the art, the present invention provides a device for preventing dangerous situations taking into account the insufficient vigilance state and the driving situation.

More particularly, according to the invention, a device for preventing dangerous situations is proposed for a driver of a transport vehicle, comprising:

a memory in which a behavioral diagram of the driver is stored during his driving, this behavioral diagram comprising data relating to at least one dangerous driving situation of the driver linked to an insufficient state of alertness of the driver,

a module for determining the driver's driving situation suitable for determining a driver's current driving situation,

- a module for estimating the driver's current state of vigilance, and

- a control unit programmed for:

- compare the driver's current driving situation with the dangerous driving situation stored in the behavior diagram,

- compare the driver's current state of vigilance with a threshold vigilance state of the behavioral pattern, and

- order the issue of an alert signal to the driver based on said two comparisons.

Thus, thanks to the invention, the driver is alerted of his alertness depending on the driving situation.

Advantageously, the control unit is programmed to control the emission of the alert signal when the current driving situation corresponds to the dangerous situation stored in the behavioral diagram, and when the current state of alertness is degraded compared to at the threshold vigilance state.

Thus, thanks to the invention, if the current driving situation of the driver is a dangerous situation, the driver is alerted when his vigilance state drops below a threshold alert state, which can for example be equal to a state satisfactory vigilance.

Advantageously, the control unit is alternatively programmed to control the emission of the alert signal when the current driving situation corresponds to the dangerous situation stored in the behavioral diagram, and that a difference between said vigilance state current of the driver and said threshold vigilance state is less than a predetermined threshold deviation.

Thus, thanks to the invention, if the driver's current driving situation is a dangerous situation, the driver is alerted when his vigilance state approaches a threshold alert state, which can for example be equal to the insufficient vigilance state of the behavioral scheme.

Advantageously, the predetermined threshold deviation is determined as a function of the driver's current driving situation.

Thus, it is possible to modulate the sensitivity of the prevention device according to the driving situation.

Other non-limiting and advantageous characteristics of the device according to the invention are as follows:

- there is further provided a driver identification system adapted to determine an identifier of the driver and in which the behavioral diagram stored in said memory is associated with this identifier of the driver;

- the dangerous driving situation and / or the threshold vigilance state, and / or the insufficient vigilance state and / or the predetermined threshold deviation is associated with said driver identifier;

said module for estimating the current vigilance state of the driver comprises a device for measuring a driver monitoring parameter and is programmed to estimate said current vigilance state for the driver as a function of the measured value of said monitoring parameter and an estimation rule;

said device for measuring the monitoring parameter comprises at least one device for capturing an image of the driver's head, and said measured monitoring parameter is determined as a function of at least one image captured from the head of the driver and is relative to the driver's gaze direction and / or to the driver's head posture, and / or to the closing of the driver's eyelids;

said device for measuring the driver monitoring parameter comprises at least one device for measuring the pressure of the hands on the steering wheel of the vehicle or a device for measuring a biometric quantity of the driver, and said measured monitoring parameter relates to hand pressure on the steering wheel of the vehicle or on a biometric quantity of the driver;

said device for measuring the driver monitoring parameter comprises at least one device for measuring a quantity relating to the driving of the vehicle by the driver and / or vehicle sensors connected to the CAN network of the vehicle, and said monitoring parameter measured relates to the driving of the vehicle by the driver and / or to data from the CAN network of the vehicle;

- the driving situation determination module comprises at least one of the following systems: a clock, a geographic position sensor, a sound sensor, a thermometer, a telephone synchronization system, a biometric sensor, and the driving situation current is determined based on at least one of the following indicators: time indicator, geographic position indicator, temperature, sound indicator, emotional indicator, circumstantial indicator;

- There is further provided a device for emitting a visual, audible, or haptic alert signal adapted to be controlled by said control unit.

The invention also provides a method for preventing dangerous situations for a driver of a transport vehicle comprising the following steps:

a) determining a behavioral pattern of the driver during his driving, this behavioral pattern comprising data relating to at least one dangerous driving situation of the driver and to a state of insufficient vigilance of the driver associated with this dangerous driving situation,

b) determine a current driving situation and estimate the driver's current alertness,

c) compare the current driving situation determined in step b) to said at least one dangerous situation of the behavioral pattern determined in step a),

d) compare the current vigilance state of the driver determined in step b) with a threshold vigilance state, and

e) order the emission of an alert signal intended for the driver according to said two comparisons carried out in steps c) and d).

Other non-limiting and advantageous characteristics of the process according to the invention are as follows:

- step a), the data relating to at least one dangerous driving situation of the driver and to a state of insufficient vigilance of the driver associated with this dangerous driving situation are recorded;

- in a step prior to step a), an identifier of the driver of the vehicle is determined, and, in step a), the predetermined behavioral pattern corresponding to the identified driver is recovered;

- the alert signal is issued in step e) when, in step c), the current driving situation corresponds to the dangerous situation stored in the behavioral diagram, and that, in step d), a deviation between said current vigilance state of the driver and said threshold vigilance state is less than a predetermined threshold deviation;

- the alert signal is issued in step e) when, in step c), the current driving situation corresponds to the dangerous situation stored in the behavioral diagram, and that, in step d), said the driver's current vigilance state is degraded compared to said threshold vigilance state;

- in step b), a driver monitoring parameter is measured and the driver's current vigilance state is estimated as a function of an estimation rule and as a function of the measured value of said monitoring parameter.

Detailed description of an exemplary embodiment

The description which follows with reference to the accompanying drawings, given by way of nonlimiting examples, will make it clear what the invention consists of and how it can be carried out.

In the accompanying drawings:

- Figure 1 schematically shows the different elements of the prevention device according to the invention;

- Figure 2 schematically shows the steps of the prevention method according to the invention; and,

- Figure 3 shows schematically the step of establishing the behavioral pattern of the driver

Device

In FIG. 1, the various elements of a prevention device 100 are shown diagrammatically.

This prevention device 100 is adapted to be mounted in a motor vehicle to alert a driver of this vehicle of possible dangerous situations which may degrade his state of alertness.

More specifically, the prevention device 100 according to the invention comprises:

a memory 30 in which a behavioral diagram of the driver is stored during his driving, this behavioral diagram comprising data relating to driving situations of the driver and to a state of alertness of the driver associated with each of these driving situations,

a module 20 for determining the driver's driving situation suitable for determining a driver's current driving situation,

- an estimation module 10 of a current vigilance state of the driver, and

- a control unit 50 programmed for:

- compare the driver's current driving situation with the dangerous driving situation stored in the behavior diagram,

- compare the driver's current vigilance state with a threshold vigilance state, and

- order the issue of an alert signal to the driver based on said two comparisons.

Here, the prevention device 100 comprises a device 200 for transmitting an alert signal suitable for being controlled by said control unit 50.

The prevention device is thus adapted to communicate with the driver via this emission device 200.

The alert signal issued is a visual, audible or even haptic signal.

In practice, the emission device 200 here comprises an existing interface of the vehicle.

More specifically, the interface transmits, at the request of the control unit 50, messages intended for the driver to warn him of dangerous situations.

The interface includes, for example, the dashboard and / or the vehicle speakers. It may also include additional elements such as a diode adapted to produce bright flashes.

The interface may also include haptic alert means arranged on the steering wheel of the vehicle, or on the seat of the driver of the vehicle. It is then adapted to emit a haptic signal, for example in the form of vibration.

Alternatively, the interface can be an independent element of the vehicle, such as a smartphone-type mobile phone or a computer connected to the vehicle or to the internet.

The messages emitted by the interface, whatever it is, are written and / or voice messages and / or in the form of light, sound or haptic signals.

In practice, the module for estimating the states of vigilance of the driver is a module known per se (more commonly called monitoring device or "Driver monitoring System" in English) adapted to estimate in real time the state of vigilance of the driver .

As shown in FIG. 1, it here comprises a measuring device 9 adapted to measure at least one driver monitoring parameter and to communicate with the control unit 50 autonomously to estimate the state of alertness of the corresponding driver to this monitoring parameter.

Here, the driver alertness estimation module 10 records information about the driver without interacting with it, so as to record the natural behavior of the driver, in the absence of any disturbance or constraint.

In the exemplary embodiment envisaged, said measuring device 9 comprises means for capturing images of the driver's head and means for analyzing these images from which the control unit 50 deduces the parameter for monitoring the driver and associated alertness.

More specifically, in the embodiment described here, the measuring device 9 comprises at least one means for capturing images of the driver's head arranged so as to capture the face of the driver in his driving position. For example, such a capture means is a camera placed behind the steering wheel of the vehicle.

The driver monitoring parameter is then measured as a function of at least one image of the driver's head captured by the capture means of the estimation module 10.

Here, the measured monitoring parameter is for example representative of the direction of the driver's gaze and / or the posture of the driver's head, and / or the closing of the driver's eyelids.

As a variant or in addition, the measuring device could include other types of sensors allowing the measurement of other types of monitoring parameter, in addition to or in replacement of the monitoring parameter already described.

In particular, the measuring device could comprise at least one device for measuring the pressure of the hands on the steering wheel of the vehicle or a device for measuring a biometric quantity of the driver so that said monitoring parameter measured by the measuring device is relating to the pressure of the hands on the steering wheel of the vehicle or to a biometric quantity of the driver such as his heart rate.

Heart rate is a monitoring parameter that can provide information on the driver’s emotional state, especially if they are in anger, stress or excitement.

It is also conceivable that the device for measuring the driver monitoring parameter comprises at least one device for measuring a quantity relating to the driving of the vehicle by the driver and / or vehicle sensors connected to a CAN network of the vehicle, so that the monitoring parameter is relative to the driving of the vehicle by the driver (acceleration and braking signals of the vehicle, information on the activation of certain functions of the vehicle, or data indicating driving along a trajectory not well suited to route) and / or data from the vehicle's CAN network.

The CAN network (by the acronym "Controller Area Network") is a multiplexed vehicle bus, which connects the various components of the vehicle and allows them to communicate with each other.

Some of the vehicle sensors connected to the vehicle CAN network include, for example, vehicle trajectory analysis means to measure a vehicle deviation parameter, the vehicle lateral stability, or even the frequency and amplitude of micro- corrections made by the driver on the steering wheel. The vigilance states estimation module could then take these driver monitoring parameters into account to determine the driver’s vigilance state.

Whatever the monitoring parameter (s) measured, the vigilance state estimation module 10 is adapted to estimate the driver's vigilance state as a function of said driver monitoring parameter (s) and the rule of estimate.

The estimation rule here includes, for example, a comparison between the measured value of the driver monitoring parameter and a threshold value of this monitoring parameter.

For example, if the driver monitoring parameter is the eyelid closing time, the driver's alertness is considered to decrease when the eyelid closing time increases.

As a variant, it can be envisaged that the estimation rule is determined by a learning algorithm adapted to establish a causal structure between the monitoring parameter and the driver's alertness state. This learning algorithm is initially trained using a predetermined initial database comprising pairs of predetermined associated input variables [monitoring parameter / vigilance state].

The learning algorithm determines for example, on the basis of this database, the probability of observing a given state of vigilance in the driver knowing that the monitoring parameter has said measured value. The estimation rule then returns the most likely state of vigilance as an output as a function of the measured value of the driver monitoring parameter.

The estimation rule includes several states of possible driver alertness at the output.

At least two states are provided, corresponding to satisfactory or insufficient vigilance.

Preferably, at least three states are provided, corresponding respectively to satisfactory, intermediate and insufficient vigilance. The intermediate and insufficient states are considered as unsatisfactory.

The vigilance state includes for example at least one level of attention or a level of alertness of the driver.

Preferably, at least three possible levels of arousal are provided: alert arousal, hypovigilance, and falling asleep.

The alert alert level corresponds to a satisfactory state of alertness on the part of the driver.

On the contrary, hypovigilance corresponds to an intermediate state of alertness and falling asleep corresponds to an insufficient state of alertness.

Hypovigilance also called relaxed arousal is a lowered level of arousal characterized by specific patterns of brain activity, in particular the presence of alpha wave whose frequency varies from 8 to 12 Hz. Body movements are more rare than in alert state of alert, the reaction time is extended. Attention is more labile, making it harder to focus on a task.

Falling asleep corresponds to an intermediate state between sleep and wakefulness characterized by specific patterns of brain activity, in particular the presence of theta wave whose frequency varies from 3.5 to 7.5 Hz, slow eye movements and a reduction in muscle tone. The presence of yawns can be detected. If the person is struggling to fall asleep, microsleeping can be observed.

The following levels of attention can be expected: concentration, low distraction, high distraction.

Concentration corresponds to a satisfactory state of alertness on the part of the driver.

On the contrary, the low distraction level corresponds to an intermediate state of alertness and the strong distraction level corresponds to an insufficient state of alertness.

The emotional states of anger, stress or excitement of the driver correspond to low (or strong) distraction levels of the driver, and therefore to an intermediate state of alertness (or insufficient).

In practice, here, the module 10 for estimating states of vigilance is adapted to transmit to memory 30 data relating to said states of vigilance.

In the exemplary embodiment described, provision is made to transmit to memory 30 the state of vigilance already estimated by the estimation module 10..

As a variant, provision could be made to transmit the monitoring parameter measured by the measuring device to the memory, the state of vigilance then being then estimated as a function of this monitoring parameter as what has been described previously.

The driving situation determination module 20 is adapted to determine the driving context, that is to say the environment in which the driver is during the journey made. This context is characterized by at least one data item called contextual data item.

Such contextual data relates for example to the time of day, the geographic position of the vehicle, the outside and / or interior temperature of the vehicle, the speed of the vehicle.

It can also be a piece of data relating to the activity of the driver at the wheel, namely the sound volume of the radio, the style of music listening, or even a piece of data revealing the fact that the driver uses his telephone: telephone conversation, manipulation of the keyboard, or observation of the screen of his telephone.

It is also conceivable that such contextual data relates to the number of passengers in the vehicle, the age of the passengers, or the sounds emitted in the vehicle: conversation between several passengers, conversation with the driver, baby crying, cries of children, etc.

We could also record data relating to the context before or after driving, which would then reveal the emotional state of the driver: important meeting before driving and generating stress, long day at work generating fatigue, shift work can disrupt the driver's circadian cycle, or even an evening of relaxation or distraction in perspective likely to generate reverie on the part of the driver.

It could still be biometric data relating to the driver’s emotional state, such as heart rate or brain activity.

In practice, each driving situation is defined by at least one of these contextual data, but more generally by a set, or a set, of data: a pair of data, a triplet of data, or more.

To determine the contextual driving data, the determination module 20 comprises at least one of the following systems: a clock, a geographic position sensor, a sound sensor, a thermometer, and a telephone synchronization system, a biometric sensor.

Each of these systems is adapted to provide contextual driving situation indicators.

In particular, the clock provides a time indicator such as the time of day, or the date of the trip.

The geographic position sensor provides a geographic position indicator, such as the vehicle's GPS coordinates. It includes a GPS here as well as road maps of the country in which the trip is made.

The sound sensor is for example a microphone which provides an audible indicator of the environment of the vehicle: conversation between passengers, crying babies, cries of children ...

The thermometer measures the temperature inside the vehicle interior, and / or the temperature outside the vehicle.

The telephone synchronization system retrieves a circumstantial indicator such as the time the driver wakes up, his possible daily schedule (meeting, evening, appointment). This synchronization system also lets you know if the driver is on the phone or manipulating or looking at the screen of his phone.

One could also consider a sensor connected to the vehicle radio and capable of providing the determination module 20 with the antenna (circumstantial indicator) and / or the listening volume of this radio (sound indicator).

The biometric sensor is a sensor adapted to record the heart rate and / or brain activity of the driver to derive an emotional indicator of the driver: tired, stressed, nervous ...

One could also consider that the system interacts with the driver before driving to question him about his emotional state and deduce said emotional indicator (tired, stressed, nervous).

Each of the sensors described above can be a dedicated sensor or a remote sensor, present in another device of the vehicle. In the latter case, the determination module 20 retrieves the indicator determined by the remote sensor on the CAN network of the vehicle.

The CAN network also allows the determination device 20 to communicate with a vehicle buffer memory in which said contextual driving data are stored in real time.

This buffer memory is adapted to communicate with the memory 30 of the prevention device 100 so that the latter receives certain data relating to the driving situation.

More specifically, provision is made here for contextual driving data sets collected over a period of time surrounding the detection of insufficient vigilance states of the driver to be stored in memory 30. The memory 30 thus stores in particular a contextual driving data set recorded in a predetermined time interval before the detection of each insufficient driver vigilance state.

Similarly, it is also expected that contextual driving data sets collected over a period of time surrounding the detection of the intermediate, respectively satisfactory, states of vigilance of the driver will be stored in the memory 30. The memory 30 thus stores in particular a data set driving context recorded in a predetermined time interval before the detection of each intermediate vigilance state, respectively satisfactory, of the driver.

The memory 30 is then adapted to associate each set of contextual data with the corresponding insufficient, intermediate or satisfactory state of vigilance to form a behavioral pattern of the driver.

In other words, here, the behavioral diagram of the driver gathers all the contextual data of the recorded driving situations having led to a state of insufficient, intermediate or satisfactory vigilance of the driver, each of these contextual data being associated with said state of insufficient, intermediate or satisfactory.

As a variant, provision could be made for the driver's behavioral diagram to gather all the contextual data of the recorded driving situations having led to an insufficient, intermediate or satisfactory state of vigilance of the driver, each of these contextual data being associated with the driver monitoring parameter measured corresponding.

In another variant, it can be envisaged that the memory includes a predetermined or pre-established behavioral diagram, that is to say already existing, of the driver associating contextual data with given states of alertness.

This pre-established behavioral pattern was for example determined on an old vehicle of the driver, or during previous journeys. This pre-established behavioral pattern is transferred to the memory of the vehicle driven by the driver, for example from another vehicle. Preferably, the behavioral pattern was established and then consolidated over several routes.

The control unit 50 is adapted to receive information from the memory 30.

It is programmed to process the behavior pattern of the driver from memory 30 in order to identify at least one dangerous driving situation, that is to say a driving situation linked to an insufficient state of alertness of the driver.

We also speak of dangerous driving situations to designate those in which an insufficient vigilance state of the driver is particularly probable.

In other words, the control unit 50 is programmed to identify driving situations with a high probability of leading the driver to an insufficient vigilance state.

In other words again, the control unit 50 is adapted to sort, in the behavioral diagram, between the contextual data being incidentally associated with an insufficient state of vigilance and the contextual data actually capable of bringing about the driver to an insufficient state of alertness.

Thus, after treatment, the behavioral diagram includes data relating to at least one dangerous driving situation of the driver linked to a state of vigilance of the insufficient driver.

To do this, the control unit 50 includes a means of statistical calculation making it possible to determine at least which contextual data of the driving situations recorded in the behavioral diagram are statistically linked to a state of insufficient vigilance of the driver.

As a variant, the control unit 50 further comprises a statistical calculation means making it possible to determine which contextual data of the driving situations recorded in the behavioral diagram are statistically linked to an intermediate state of vigilance of the driver.

This statistical calculation means comprises, for example, a conditional probability and / or statistical likelihood calculation means and / or a correlation calculation means.

The control unit 50 is thus adapted to identify, by a conditional probability calculation and / or statistical likelihood and / or a correlation calculation, the contextual data of each driving situation of the behavioral diagram actually linked to the state of insufficient vigilance associated with this driving situation, these identified contextual data defining one or more new situations hereinafter called "dangerous situations".

Thus, after treatment, for each dangerous situation corresponds at least one contextual datum of the behavioral diagram, but more generally a set, or a set, of data: a data pair, a data triplet, or more.

In practice, certain contextual data alone suffices to define a dangerous situation. For example, the time of day is a contextual data very indicative of the circadian cycle and allows to define by itself a dangerous situation when the time of day is such that it is associated with the insufficient state of vigilance of the driver. At this time of day the driver is drowsy.

On the contrary, certain contextual data alone are not sufficient to define a dangerous situation. This is the case, for example, of the speed variation of the vehicle.

Preferably, each dangerous situation is associated with at least two contextual data. Advantageously, associating a dangerous situation with two contextual data makes it possible to specify dangerous situations and therefore to limit the number of situations considered to be dangerous in an unjustified manner.

The set of data defining a dangerous situation is not necessarily the same as the set of data defining any situation in the behavioral diagram stored in memory 30.

In other words, there is not necessarily a strict equivalence between a driving situation recorded in the behavioral diagram of the memory 30 and a driving situation identified as dangerous by the control unit 50, said control unit command 50 being adapted to separately evaluate the conditional probability or the likelihood or the correlation of each datum of the contextual data set of the behavioral diagram, and the conditional probability or likelihood or correlation of each subset of data (data pair, triplet of data ...) of the contextual data set of the behavioral diagram, associated with an insufficient state of vigilance.

For example, the control unit 50 can determine, by calculating conditional probability, likelihood or correlation, that an insufficient state of alertness corresponding to the level of falling asleep is statistically probable at a given time of the day, for example at 7 p.m. when the driver comes home from work. It can for example determine that the insufficient vigilance state corresponding to the level of sleep is likely at 2 p.m., when the interior temperature exceeds 25 degrees. It can also determine that an insufficient state of alertness corresponding to a high level of distraction is observed at 6 p.m. and, possibly, associate this with the fact that the driver phones while driving when he takes his vehicle to go home or the fact that children are present on board.

On the other hand, when a situation has been identified as dangerous for a large number of drivers, it can be considered that the situation is dangerous for all drivers. Thus, if 60% of drivers are found to be in a dangerous situation after dark, then this situation can be considered dangerous for all drivers.

The control unit 50 is programmed to, when the driver's current driving situation corresponds to an identified dangerous situation, alert the driver of the vehicle to its alert state as soon as the latter is close to the threshold alert state.

In practice, the threshold vigilance state is determined according to the insufficient vigilance state of the behavioral pattern.

In particular, the threshold vigilance state is a state in which the driver's vigilance is greater than or equal to that of the driver's insufficient vigilance state.

The threshold alert state can also be determined based on the driver's current driving situation.

In practice, the driving situation determination module 20 is adapted to determine the driver's current driving situation, that is to say the driving situation in real time, and the control unit 50 is programmed to compare this current driving situation to said at least one dangerous situation identified.

More specifically, the control unit 50 is programmed to estimate that the current driving situation corresponds to at least one of the dangerous situations of the behavioral diagram when the contextual data defining the dangerous situation are reproduced in the current driving situation.

Similarly, the driver vigilance state estimation module 10 is adapted to estimate the driver's current vigilance state, that is to say his driving state in real time, and the control unit 50 is adapted to compare the driver's current vigilance state with said threshold vigilance state.

More specifically, the control unit 50 is adapted to detect if the current vigilance state is degraded with respect to said threshold vigilance state.

As an example, the control unit 50 can be programmed to determine the threshold vigilance state as being equal to the satisfactory vigilance state.

In a first example, it is considered that the prevention device is capable of detecting three different vigilance states: the insufficient vigilance state, the intermediate vigilance state and the satisfactory vigilance state.

In this case, the control unit 50 controls the emission of the alert signal as soon as the current situation is dangerous, and the driver's current alert state is degraded compared to the satisfactory alert state, c '' to say as soon as its state of vigilance is intermediate, and a fortiori when it is insufficient.

In another example, it could be envisaged that the prevention device 1 is capable of detecting ten different states of vigilance, classified as the least satisfactory state of vigilance, that is to say the state of vigilance "insufficient". (state of vigilance number 1) to the most satisfactory state of vigilance (state of vigilance number 10).

In this second example, it can be envisaged that the threshold vigilance state is state of vigilance number 3.

The control unit then compares the current vigilance state and the vigilance number 3 state, and triggers the alert signal as soon as the current vigilance state is less than or equal to the vigilance state number 3.

In this case, the driver is alerted if he is in one of the alert states number 1,2 or 3.

In summary, as soon as the current vigilance state is degraded compared to the threshold vigilance state, the control unit 50 is adapted to emit the alert signal intended for the driver of the vehicle.

Thus, the control unit 50 is programmed to control the emission of the signal when the current driving situation corresponds to the dangerous situation stored in the behavioral diagram, and when the current vigilance state is degraded compared to the state threshold vigilance.

According to a variant of the invention, the control unit 50 is adapted to evaluate the difference between the driver's current vigilance state and the threshold vigilance state, and to determine if this difference is less than a threshold difference predetermined.

For example, the control unit here determines that the threshold vigilance state is equal to the insufficient vigilance state.

In practice, to evaluate the difference between the current and threshold vigilance states, the control unit 50 associates with each vigilance state a digital value and compares these digital values.

In an example where the prevention device is capable of detecting three different vigilance states, the control unit associates for example with the insufficient vigilance state (here equal to the threshold vigilance state), the digital value 0, in the intermediate vigilance state, the numerical value 1, and in the satisfactory vigilance state, the numerical value 2.

In this case, the control unit determines that the predetermined threshold deviation is equal to 1. Thus, by comparing the current and insufficient vigilance states, that is to say by calculating the difference, in absolute value, between the values numerical associated with said current and insufficient vigilance states, the control unit 50 commands the emission of the alert signal as soon as said difference is less than or equal to 1, that is to say as soon as the vigilance state current is intermediate or insufficient.

Alternatively, other numerical values could be associated with the states of insufficient, intermediate and satisfactory alertness.

In another variant, the threshold vigilance state could be associated with a specific numerical value, which is not equal to that associated with the insufficient vigilance state.

In another example where the prevention device is capable of detecting ten different vigilance states, classified as the least satisfactory state of vigilance, that is to say the "insufficient" state of vigilance with which the digital value 1, in the most satisfactory state of vigilance with which the digital value 10 is associated, it is possible that the control unit can evaluate the difference between the current state of vigilance and state of vigilance number 3, associated with the numerical value 3.

The control unit then triggers the alert signal when this deviation is less than or equal to a predetermined threshold deviation, for example equal to 2. In this case, the driver is alerted if he is in one of the vigilance states number

1,2, 3, 4 or 5, associated with the numerical values 1,2,3, 4, and 5.

In another variant, the difference between the driver's current vigilance state and the threshold vigilance state could be evaluated by calculating the algebraic (and not absolute) difference between the values associated with said threshold and current vigilance states and by determining the value of the appropriate predetermined threshold deviation.

As a variant, to evaluate the difference between the current and threshold vigilance states, the control unit can for example evaluate the difference between the monitoring parameter or parameters corresponding to said current and threshold vigilance states.

For example, when the threshold vigilance state is equal to the insufficient vigilance state, the control unit evaluates the difference between the monitoring parameter or parameters corresponding to said current and insufficient vigilance states.

According to this variant, there are then as many predetermined threshold deviations as there are types of monitoring parameters.

However, the predetermined threshold difference is for example chosen equal to the difference existing between the value of the monitoring parameter associated with the satisfactory vigilance state and the value of this same monitoring parameter associated with the threshold vigilance state. (or insufficient when the threshold and insufficient vigilance states are identical).

The control unit could also be programmed to determine the predetermined threshold deviation so that it is equal to the difference existing between the value of the monitoring parameter associated with the intermediate vigilance state and the value of this same parameter. monitoring associated with the threshold vigilance state (or insufficient when the threshold and insufficient vigilance states are identical).

In summary, as soon as the difference between the driver's current vigilance state and the threshold vigilance state is less than said predetermined threshold deviation, the control unit 50 is adapted to emit the alert signal intended for the driver of the vehicle.

Thus, the control unit 50 is programmed to control the emission of the signal when the current driving situation corresponds to the dangerous situation stored in the behavioral diagram, and when the difference between said current vigilance state of the driver and the threshold vigilance state is less than said predetermined threshold deviation.

Advantageously, thanks to the invention, whatever the variant embodiment, the driver is alerted of his current state of alertness although this is not necessarily insufficient. Indeed, the control unit 50 having determined that the current driving situation is dangerous alerts the driver to his state of alertness as he is close to being insufficient.

In other words still, thanks to the invention, the sensitivity of the device for transmitting the alert signal 200 is improved when the current driving situation is a dangerous driving situation.

For example, when the driver's current driving situation is dangerous, the control unit 50 alerts the driver when his current alert state is intermediate.

On the other hand, when the driver's current driving situation is not dangerous, the control unit 50 only alerts the driver to his current alert state when the latter is insufficient.

To do this, the control unit 50 can determine the threshold vigilance state as a function of the current driving situation: if it is a dangerous situation, the threshold vigilance state can be closer to the state of vigilance satisfactory only when the current driving situation is not identified as dangerous.

Similarly, according to the variant embodiment, the control unit 50 can determine the predetermined threshold deviation as a function of the current driving situation: if it is a dangerous situation, the predetermined threshold deviation can be more great only when the current driving situation is not identified as dangerous.

The threshold vigilance state and / or the predetermined threshold deviation can still be different for each dangerous situation. In other words, the threshold alert state and / or the predetermined threshold deviation may depend on the dangerous driving situation identified.

In addition, the control unit 50 can also be programmed to send a message to the driver in order to warn him of a dangerous situation.

In particular, the control unit 50 is then adapted to inform the driver of the dangerous situations identified so that he becomes aware of them.

More specifically, the control unit 50 is adapted to communicate with the transmission device 200 (here directly with the interface) so that the latter transmits an intelligible message to the driver.

The emission device 200 can warn it at any time, preferably when the vehicle engine is off, or when the vehicle is stopped.

The informative message sent can be more or less complex. It includes at least the information that a dangerous situation is observed when a specific contextual situation indicator has a given value.

In addition, the control unit 50 can also be adapted to send a message to the driver based on the driver's current driving situation, regardless of the driver's estimated alertness.

To do this, in practice, the driving situation determination module 20 is adapted to determine the current driving situation of the driver, and the control unit 50 is programmed to compare this current driving situation with said at least one dangerous situation identified in the behavioral diagram.

The control unit 50 is then adapted to send the message to the driver as a function of this comparison, preferably solely as a function of this comparison.

Of course, the device 200 for emitting the alert signal is also suitable for emitting the alert signal to the driver when the driver's current alert state is insufficient, whatever the driving situation in progress. In this case, the alert signal is preferably an audible signal.

In other words, it is possible to choose that the emitting device can emit an alert signal only as a function of the driver's alertness state.

Here, the prevention device 100 further comprises an identification system 40 for the driver (see FIG. 1).

This driver identification system 40 is suitable for determining a driver identifier.

In other words, it is suitable for associating the driver of the vehicle with an identifier.

Advantageously, this identification system 10 allows the prevention device 100 according to the invention to be adaptable to several drivers of the same vehicle.

Here, the identification system 40 is a system by facial recognition of the driver.

Alternatively, it could be envisaged that it is a system by fingerprint recognition, by recognition of a specific driver key, or more simply by entering an authentication code (type PIN code) . This last option requires the intervention of the driver, before setting the vehicle in motion.

The identification system 40 is adapted to communicate with the memory 30 and the control unit 50 of the prevention device 100 using the identifier.

More precisely, the behavioral diagram stored in the memory 30 is associated with this identifier of the driver so that it is specific to this driver in particular.

The memory 30 can thus be here adapted to store several behavioral patterns, and the control unit 50 is adapted to process the behavioral pattern associated with the identified driver.

When the device according to the invention comprises the identification system 40, provision can also be made for the dangerous driving situation and / or the insufficient vigilance state and / or the predetermined threshold deviation to be associated with said driver identifier. These parameters can then be personalized according to the driver.

In a variant, the control unit of the prevention device according to the invention can also be programmed to further identify, from said behavioral diagram, at least one potentially dangerous situation correlated with a state of alertness of the intermediate driver, and emit a message to the driver to warn him of this potentially dangerous situation.

To this end, the control unit performs a statistical calculation to determine the contextual data of the driving situations recorded statistically associated with an intermediate state of vigilance of the driver.

Process

The following section describes the process for preventing dangerous situations for a driver of a transport vehicle according to the invention.

This prevention method is implemented by the prevention device 100 according to the invention described above.

In practice, the prevention method according to the invention comprises the following steps:

a) determining a behavioral pattern of the driver during his driving, this behavioral pattern comprising data relating to at least one dangerous driving situation of the driver and to a state of insufficient vigilance of the driver associated with this dangerous driving situation,

b) determine a current driving situation and estimate the driver's current alertness,

c) compare the current driving situation determined in step b) to said at least one dangerous situation of the behavioral pattern determined in step a),

d) compare the current vigilance state of the driver determined in step b) with a threshold vigilance state, and

e) order the emission of an alert signal intended for the driver according to said two comparisons carried out in steps c) and d).

Step a)

In a first simplified embodiment of step a), provision is made to load into memory 30 of the prevention device 100, the behavioral diagram of the preset driver.

To this end, the identification system 40 identifies the driver, determines his identifier and transmits it to memory 30 so that it loads the behavior pattern of the driver associated with this identifier.

This first embodiment is implemented in particular when the behavioral pattern of the driver has been established during a prior step of recording this behavioral pattern.

Advantageously, this behavioral pattern can be transferred from one vehicle to another.

In a second embodiment, it is planned to record, in step a), the behavioral diagram of the driver by recording in the memory 30 of the transport vehicle the contextual data relating to the driving situations associated with the vigilance states of the driver.

In other words, while the driver is driving the vehicle, data relating to driving situations and the estimated alertness of the corresponding driver are recorded.

This data is for example recorded at regular time intervals, for example every minute.

Here, for example, provision is made to record in the memory 30 all the contextual data relating to the driving situations as well as the weakest state of alertness detected during the minute surrounding each driving situation.

In particular, but not only, contextual data relating to driving situations surrounding the detection of a significant event, for example an insufficient vigilance state, is recorded.

For example, the contextual data associated with the driver's insufficient vigilance states, namely corresponding to the levels of drowsiness and / or strong distraction of the driver, are recorded in the memory 30.

The establishment of the behavioral pattern requires the achievement of a sufficient number of journeys in order to accumulate enough contextual data associated with each state of vigilance: insufficient, intermediate, satisfactory.

Thus, the detection of dangerous situations which are not in reality, also called "false positives", will be reduced.

For example, it is estimated that, if a particular insufficient state of vigilance occurs daily, the behavioral pattern acquired after a period greater than one month, preferably greater than three months, will make it possible to implement the method according to the invention. After three months of daily journeys, the behavioral diagram stored in memory 30 will be considered very reliable.

On the other hand, if it is detected in 60% of drivers that nightfall is associated with an insufficient state of alertness, then this contextual data ("nightfall") will define a situation considered to be dangerous for all. drivers, that is, in all behavioral patterns.

There is shown schematically in FIG. 3 the steps for establishing the behavioral diagram of the driver according to the second embodiment.

In a first step G1, the estimation 10 and determination modules 20 of the prevention device 100 determine the data relating to the state of alertness of the driver at time t as well as the data relating to the driving situation at this time. instant t.

In other words, said modules 10, 20 respectively determine the driver's current vigilance state and the associated current contextual data.

In a second step G2, these data are stored in a buffer memory.

In parallel with this second step G2, the control unit 50 analyzes in a third step G3 whether the state of alertness of the driver detected at time t is an insufficient, intermediate or satisfactory state of alertness.

In practice, the driver's vigilance state is determined here using the driver monitoring parameter and the estimation rule.

When the estimated vigilance state shows that the driver's vigilance is insufficient, the contextual data recorded in the buffer memory for a given time interval preceding this insufficient vigilance state are stored in memory 30 during a fourth step G4 . These contextual data are associated in the memory 30 with the vigilance state, either directly by being associated with the digital value corresponding to said vigilance state, or indirectly by being associated with said corresponding measured monitoring parameter.

In practice, the memory 30 stores for example the data relating to the driving situation at the instant t of the detection of the insufficient vigilance state, as well as the contextual data relating to the driving situations established every minute until 5 minutes before time t.

Similarly, when the estimated state of vigilance shows that the driver's vigilance is intermediate or satisfactory, the contextual data recorded in the buffer memory for a given time interval preceding this intermediate or satisfactory state of vigilance are stored in memory 30 during of the fourth step G4. These contextual data are associated in the memory 30 with said measured monitoring parameter.

The couple formed by these contextual data and the numerical value associated with the estimated state of alertness contributes to establishing the behavioral pattern of the driver.

As a variant, the couple formed by these contextual data and the measured monitoring parameter or the estimated state of alertness contributes to establishing the behavioral pattern of the driver.

By repeating steps G1 to G4 at different times t, the behavioral pattern is established step by step.

As a variant, as mentioned previously, it is possible to store in the memory only the data relating to the situation at the instant t of the detection of the insufficient, intermediate and satisfactory state of vigilance, or to store the contextual data surrounding the instant t, that is to say anterior and posterior to time t.

In practice, the identification system 40 identifies the driver, determines his identifier and transmits it to the memory 30 so that the latter stores in memory each data item of the behavioral diagram corresponding to the identifier of the driver.

It is thus possible to store the behavioral diagrams of different conductors in the same memory 30.

In addition, advantageously, the behavioral diagram is preferably updated as and when journeys made with the vehicle. Thus, the behavioral pattern takes into account possible changes in the driving habits of the driver.

As a variant, it could be envisaged that the behavioral pattern should be established once and for all, after a determined number of journeys, or after a chosen cumulative driving time.

Of course, it is possible to combine the first and second embodiments so that the behavioral pattern of the driver can include a pre-established part in a step prior to step a), and another part recorded over time according to the second embodiment.

This allows in particular to accumulate a large amount of data and to improve the behavioral pattern over time, including when the driver changes vehicle.

The control unit 50 retrieves from memory 30 the behavioral diagram established in step a) so as to process this behavioral diagram.

With this processing, the control unit identifies at least one dangerous situation, that is to say a situation correlated with an insufficient vigilance state of the driver.

After treatment, the behavioral pattern then includes data relating to at least one dangerous driving situation of the driver and an insufficient state of vigilance of the driver.

The dangerous situation is identified here by a statistical calculation taking into account the data relating to the driver's driving situations and the driver's vigilance states insufficient in the behavioral diagram.

More precisely, the statistical calculation means of the control unit 50 implements a conditional probability calculation taking into account the contextual data of the driving situations recorded in the behavioral diagram of the memory 30, and the data relating to the states of insufficient vigilance stored in memory 30.

This conditional probability calculation can be any conditional probability calculation, in particular of statistical likelihood.

This involves, for example, determining the probability that the driver has an insufficient state of alertness knowing a precise contextual data set.

Statistical likelihood is understood here as the probability that an insufficient state of alertness occurs in a given driving situation.

As a variant, the statistical calculation means of the control unit 50 can implement a correlation calculation between the contextual data of the driving situations recorded in the behavioral diagram of the memory 30 and the data relating to the vigilance states recorded in memory 30.

This correlation calculation can be any conventional correlation calculation providing a correlation coefficient. For example, it can be a Pearson correlation calculation, a partial correlation calculation, a Spearman correlation calculation, or a Kendall Tau.

In the case of Pearson correlation, if the correlation coefficient is 1, in absolute value, then the two data are perfectly correlated. The closer the correlation coefficient to 1, the stronger the correlation between the contextual data and the state of vigilance. Conversely, a correlation coefficient equal to 0 means that the contextual data are not correlated with the vigilance state.

Here, it is considered for example that a contextual data item is correlated with an insufficient vigilance state if the correlation coefficient is greater than or equal to 0.6 between said contextual item data and the monitoring parameter corresponding to this insufficient state.

We can consider that a machine learning algorithm and / or probability calculations from the conditional probabilities for each contextual indicator can estimate the likelihood of a dangerous situation. Then, for example, if this likelihood is greater than 0.7, the dangerous situation is recorded as such in the behavioral diagram.

The contextual data correlated with an insufficient state of vigilance then defines a dangerous situation.

The statistical calculation of conditional probability and / or statistical likelihood and / or correlation can also be performed between a pair of contextual data and the monitoring parameter corresponding to an insufficient state of vigilance, or even between a subset of contextual data (data triplet or more) and said monitoring parameter. This calculation then identifies more complex dangerous situations.

Step b)

In step b), the driving situation determination module 20 determines the driver's current driving situation, and the alertness estimation module 10 estimates the driver's current alertness state.

To do this, the driving situation determination module 20 monitors the driver's current driving situation, that is to say the contextual data at given times t of a journey.

Similarly, the driver vigilance state estimation module 10 monitors the driver's current vigilance state, that is to say the monitoring parameters at given times t of a journey.

More specifically, in step b), a driver monitoring parameter is measured and the driver's current vigilance state is estimated as a function of the estimation rule chosen and as a function of the measured value of said monitoring parameter.

Step c)

In step c), the control unit 50 compares the current driving situation with at least one dangerous situation identified in the behavior diagram.

In practice, the control unit 50 compares this contextual data defining the current driving situation of the journey with that defining each of the dangerous situations identified in step b).

If some of the contextual data detected during the journey corresponds to that defining a dangerous situation, the control unit 50 goes to step d).

If, on the contrary, no contextual data detected defining the current driving situation of the driver does not correspond to at least one of the contextual data defining a dangerous situation of the behavioral diagram, the control unit 50 repeats steps b) and c).

Step d)

In step d), the control unit 50 compares the current vigilance state with the threshold vigilance state.

In practice, the control unit 50 evaluates whether said driver's current vigilance state is degraded compared to said threshold vigilance state.

If the driver's current alert state is degraded or equal to the threshold alert state, the control unit 50 goes to step e).

If, on the contrary, the current vigilance state is greater than the threshold vigilance state, the control unit 50 repeats steps b) and c).

Advantageously, the threshold vigilance state can be adjusted by the driver so that he can choose the vigilance states in which he wishes to be alerted.

For example, the driver can choose to be alerted only of the vigilance states degraded compared to the threshold vigilance state, or of the degraded vigilance states or equal to the threshold vigilance state, or of the vigilance states either close to the threshold vigilance state but greater than the latter, either degraded or equal to said threshold vigilance state.

As a variant, the control unit 50 evaluates the difference between the driver's current vigilance state and the threshold vigilance state, and determines whether this difference is less than the predetermined threshold difference, as explained above. .

If the difference between the driver's current vigilance state and the threshold vigilance state is less than the predetermined threshold deviation, the control unit 50 goes to step e).

If, on the contrary, the difference between the driver's current vigilance state and the threshold vigilance state is greater than or equal to the predetermined threshold difference, the control unit 50 repeats steps b) and c).

Advantageously, the predetermined threshold deviation can be adjusted by the driver so that he can choose the vigilance states in which he wishes to be alerted.

For example, the predetermined threshold deviation is adjusted by the driver so that he can choose to be alerted of all intermediate vigilance states or only of those being almost insufficient.

Step e)

In step e), the control unit 50 controls the emission of the alert signal intended for the driver.

More specifically, the control unit 50 commands the device 200 for emitting the alert signal to emit said alert signal.

Thus, the driver is warned that his vigilance state is degraded compared to the threshold alert state, which may in particular be equal to the satisfactory alert state, or that he is approaching the threshold alert state. , which can in particular be equal to the state of insufficient vigilance.

In other words, when the current driving situation is considered to be dangerous, the control unit 50 is programmed to alert the driver before his state of alertness really becomes insufficient.

In summary, the alert signal is issued in step e) when:

- in step c), the current driving situation corresponds to the dangerous situation stored in the behavioral diagram, and that

- in step d), said current vigilance state of the driver is degraded compared to said threshold vigilance state.

Alternatively, the alert signal is issued in step e) when:

- in step c), the current driving situation corresponds to at least one of the dangerous situations stored in the behavioral diagram, and that,

- in step d), the difference between said current vigilance state of the driver and said threshold vigilance state is less than said predetermined threshold deviation.

In addition, the control unit 50 can also warn the driver of the vehicle of the dangerous situations identified in the behavior diagram.

For example, the control unit 50 informs the driver of all the contextual data defining dangerous situations.

To do this, the control unit 50 sends a message to the driver, for example when the vehicle engine is off, or when the vehicle is stopped.

In addition, thanks to steps b) and c) of the method according to the invention, provision can be made for the control unit 50 to send messages as a function of the dangerous driving situations actually encountered during a journey.

To do this, if some of the contextual data detected during the trip corresponds to that defining a dangerous situation, the control unit sends a message to the driver to warn him of this dangerous situation.

In other words, if some of the contextual data corresponding to the current driving situation are those defining a situation identified as dangerous by the control unit 50, the latter warns the driver, regardless of the driver's current alertness .

For example, if the probability of the occurrence of an insufficient vigilance state of the driver according to a given driving situation is greater than 10, it is likely that this driving situation is dangerous and leads to a deterioration of the vigilance. In this case, an interrogative message can be sent to the driver of the vehicle in order to question him about his current state of alertness.

If the probability of a deterioration in alertness depending on a given driving situation is greater than 20, then it is very likely that this driving situation is dangerous and leads to a deterioration in alertness. . In this case, a preventive message can be sent to the driver, such as "be careful in this driving situation".

If the probability of a deterioration in alertness according to a given driving situation is greater than 40, then it is extremely likely that this driving situation is dangerous and leads to a deterioration in alertness . In this case, an alert message can be sent directly to the driver, without waiting for the vehicle to stop, and regardless of the driver's current alertness.

The message can be issued at any time from the identification of the current situation as a dangerous situation.

For example, the message can be transmitted immediately upon identification of the dangerous situation. According to another example, the message can be transmitted only at the end of the journey, as a summary message of the dangerous situations encountered during the journey.

For this addition, the control unit 50 emits two types of messages: informative messages and advice.

Informative messages are for example sent at the end of a journey. Such messages allow the driver to avoid, in the future, the dangerous situations in which he found himself during the journey.

Thus, these informative messages are intended to help the driver become aware of his driving habits so that he changes them so as to reduce the number of dangerous situations in which he is likely to find himself.

Such informative messages could for example be: "Nightfall is a dangerous situation", or "The hour after the lunch break is a dangerous situation".

Advantageously, the informative messages can also be personalized so as to have more impact on the driver. For example, such personalized informational messages could be "We have found that darkness has a significant impact on your driving and is a dangerous situation for you".

The advice is intended to offer an alternative to the driver so that he changes his current and dangerous driving situation to a non-dangerous driving situation.

As a variant or in addition, advice can also be given to prevent the dangerous situation from occurring.

Advantageously, the advice allows the driver to avoid the dangerous situation and, consequently, to prevent his state of alertness from deteriorating until it becomes insufficient.

They are preferably issued as soon as the dangerous situation is detected.

Such advice can for example be "Take a break", or even "Lower the volume of the radio to preserve your concentration behind the wheel".

In some cases, the message can also be issued before the dangerous situation occurs. This is particularly the case when the dangerous situation is defined by a given time or geographic position indicator: the control unit is then programmed to send the message when the time of the current day determined is within a given time range before the time of day identified as associated with a dangerous situation or to send the message when the vehicle's GPS position is located within a given distance from the geographic location associated with a dangerous situation.

In general, the time of issue of the message can thus be determined according to the dangerous situation identified.

Such advice given before the current situation is dangerous can for example be "Take a break before dark", or "Keep your eyes on the road".

Advantageously, the prevention method according to the invention warns the driver of certain dangerous driving situations, in which he is or may have been, independently of the assessment of states of alertness.

Also in addition, the control unit of the prevention device according to the invention identifies, from said behavioral diagram, at least one potentially dangerous situation linked to a state of alertness of the intermediate driver, and sends a message to the driver to warn him of this potentially dangerous situation.

In addition, conventionally, when the current vigilance state estimated by the estimation module 10 is insufficient, the control unit 50 is programmed to emit an alert signal intended for the driver of the vehicle, whatever or the driving situation. An alarm can thus also be issued when no dangerous situation is detected. The prevention system, however, has increased sensitivity when a dangerous situation is detected.

The alert signal issued in this supplement can be an audible or visual signal and include a voice message.

The prevention device according to the invention and the associated prevention method can also be implemented in transport vehicles such as for example trucks, trains, barges or other boats, trams or even buses.

As has already been introduced previously, the module for estimating the device according to the invention can be adapted to evaluate states of alertness additional to those described above, namely additional to the states of alertness satisfactory, intermediate and insufficient. For example, there could be several intermediate states of vigilance: a quasi-satisfactory state, an average state, a quasi-insufficient state.

We could then consider defining several types of dangerous situations, namely fairly dangerous situations associated with near-insufficient vigilance states, moderately dangerous situations associated with average vigilance states, and less dangerous situations associated with quasi-vigilance states. satisfactory.

In practice, the threshold vigilance state can be determined according to the current driving situation. Thus, it may be closer to the satisfactory alert state when the driving situation is dangerous, and closer to the insufficient alert state when the driving situation is not dangerous.

Similarly, the predetermined threshold deviation for comparing the driver's current alert state and the threshold alert state will be determined according to the current driving situation. Thus, the predetermined threshold deviation may be greater in the case of a very dangerous situation than in the case of a moderately dangerous or little dangerous situation. Similarly, the predetermined threshold deviation may be greater in the case of a moderately dangerous situation than in the case of a not very dangerous situation.

Claims (14)

1. Device for preventing (100) dangerous situations for a driver of a transport vehicle, comprising: a memory (30) in which a behavioral diagram of the driver is stored during his driving, this behavioral diagram comprising data relating to at least one dangerous driving situation of the driver linked to an insufficient state of alertness of the driver, a module for determining (20) the driving situation of the driver adapted to determine a current driving situation of the driver, - an estimation module (10) of a current vigilance state of the driver, and - a control unit (50) programmed for: - compare the driver's current driving situation with the dangerous driving situation stored in the behavior diagram, - compare the driver's current vigilance state with a threshold vigilance state, and - order the issue of an alert signal to the driver based on said two comparisons. 2. Prevention device (100) according to claim 1, in which the control unit (50) is programmed to control the emission of the alert signal when the current driving situation corresponds to the dangerous situation stored in the diagram. behavioral, and that the current vigilance state is degraded compared to said threshold vigilance state. 3 prevention device (100) according to claim 1, wherein the control unit (50) is programmed to control the emission of the alert signal when the current driving situation corresponds to the dangerous situation stored in the behavioral diagram , and that a difference between said current vigilance state of the driver and said threshold vigilance state is less than a predetermined threshold difference. 4. A prevention device according to claim 3, in which the predetermined threshold deviation is determined as a function of the driver's current driving situation. 5. Prevention device (100) according to one of claims 1 to 4, in which there is further provided an identification system (40) of the driver adapted to determine an identifier of the driver and in which the behavioral pattern stored in said memory is associated with this driver identifier. 6. Prevention device according to one of claims 1 to 5, in which said module for estimating (10) the driver's current state of vigilance comprises a device (9) for measuring a driver monitoring parameter and is programmed to estimate said current alert state of the driver based on the measured value of said monitoring parameter and an estimation rule. 7. A prevention device according to claim 6, in which said device for measuring (9) the monitoring parameter comprises at least one device for capturing an image of the driver's head, and said measured monitoring parameter is determined as a function of 'at least one captured image of the driver's head and relates to the driver's gaze direction and / or the posture of the driver's head, and / or the closing of the driver's eyelids. 8. Prevention device according to one of claims 1 to 7, in which the module for determining (20) the driving situation comprises at least one of the following systems: a clock, a geographic position sensor, a sound sensor , a thermometer, a telephone synchronization system, a biometric sensor, and the current driving situation is determined based on at least one of the following indicators: time indicator, geographic position indicator, temperature, sound indicator, emotional indicator, indicator circumstantial. 9. Prevention device according to one of claims 1 to 8, in which there is further provided a device for transmitting (200) a visual, sound or haptic alert signal adapted to be controlled by said unit. control (50). 10. Method for preventing dangerous situations for a driver of a transport vehicle comprising the following steps: a) determining a behavioral pattern of the driver during his driving, this behavioral pattern comprising data relating to at least one dangerous driving situation of the driver and to a state of insufficient vigilance of the driver associated with this dangerous driving situation, b) determine a current driving situation and estimate the driver's current alertness, c) compare the current driving situation determined in step b) to said at least one dangerous situation of the behavioral pattern determined in step a), d) compare the current vigilance state of the driver determined in step b) with a threshold vigilance state, and e) order the emission of an alert signal intended for the driver according to said two comparisons carried out in steps c) and d). 11. The prevention method according to claim 10, according to which, in step a), the data relating to at least one dangerous driving situation of the driver and to at least one state of insufficient vigilance of the driver associated with this situation are recorded. dangerous driving. 12. Prevention method according to one of claims 10 and 11, according to which the alert signal is emitted in step e) when: - in step c), the current driving situation corresponds to the dangerous situation stored in the behavioral diagram, and that - in step d), said current vigilance state of the driver is degraded compared to said threshold vigilance state. 13. A prevention method according to one of claims 10 to 12, according to which the alert signal is issued in step e) when: - in step c), the current driving situation corresponds to the dangerous situation stored in the behavioral diagram, and that, - in step d), a difference between said current vigilance state of the driver and said threshold vigilance state is less than a predetermined threshold difference. 14. Prevention method according to one of claims 10 to 12, according to which, in step b), a driver monitoring parameter is measured and the driver's current vigilance state is estimated according to a rule estimation and as a function of the measured value of said monitoring parameter.