METHOD, DEVICE AND COMPUTER PROGRAM PRODUCT FOR

ESTIMATING THE TIREDNESS OF A MOTOR VEHICLES DRIVER AND A

MOTOR VEHICLE INCLUDING SUCH A DEVICE

TECHNICAL FIELD

The present invention relates generally to the realm of vehicles and the detection of fatigue in drivers of such vehicles. In particular, the invention relates to a method, a device and a computer programme product for assessing the tiredness of a driver of a motor vehicle, and to a motor vehicle comprising such a device.

STATE OF THE ART

Within the realm of vehicles and, in particular, within the area of transport vehicles it is important that the vehicle driver be wide-awake.

It is therefore desirable to be able to assess whether the driver is wide-awake or has various degrees of fatigue. A tired driver can then be informed in various ways so that various measures can be adopted.

There are a number of different procedures for assessing fatigue. Some of them are based on the driver being monitored to determine where he/she is tired or not. Such a method is for example described in US 6,154,123, in which visual and acoustic response, mechanical response and steering wheel position are received from a driver to determine how fresh he/she is. A similar solution is described in US 6,313,749.

Many such procedures, however, are only adapted to detecting various aspects of the vehicle's behaviour and using them to determine a driver's tiredness.

For example, GB 2 346 998 describes how a pattern of variation in the force exerted on a steering wheel is stored during a first period of time. This pattern of variation is thereafter compared with subsequent measured values, and if the difference between

the values exceeds a statistically defined threshold a warning signal is generated. WO 2005/124713 describes a method in which changes in the steering wheel angle over time are analysed and evaluated with respect to control activity time interval and control break time interval, with particular attention to the length of break time intervals.

US 5,900,819 describes a system which measures the vehicle's speed and various magnitudes from the vehicle's axles, e.g. lateral acceleration. These magnitudes are compared thereafter with thresholds, and a fatigue counter is incremented when these thresholds are exceeded. There is also a cycle counter which is incremented each time a comparison of all the magnitudes is effected. A quotient between these counter values is thereafter compared with a fatigue threshold, and tiredness is indicated if this threshold is exceeded.

US 6,756,903 describes another system, which measures a number of different physical magnitudes such as vehicle speed, steering wheel angle, angular velocity, angular acceleration and lateral acceleration. These are thereafter divided into driver- initiated movements and non-driver-initiated movements. The magnitudes in the two categories are thereafter put into respective models, viz. a statistical model for driver- initiated movements and a model for non-driver-initiated movements. From these models a value for driver-initiated movements and a value for non-driver-initiated movements are thus obtained. A wakefulness index is thereafter obtained by dividing these values by one another.

There are thus a number of different ways of determining fatigue, but many different magnitudes usable for detecting it. Some of these magnitudes are perhaps not always available or not always calculable, so the model used is in principle usually general, which means that there are situations where the model used for assessing fatigue is not optimum. This may lead to fatigue being indicated in cases where it should not, which may be regarded as annoying to the driver. It may also mean that fatigue is not indicated when a driver really is tired, which may have fatal consequences.

There is therefore a need to improve fatigue detection in motor vehicles.

SUMMARY OF THE INVENTION

An object of the present invention is to provide improved fatigue detection in relation to a driver of a motor vehicle.

According to an aspect of the present invention, this object is achieved by a procedure for assessing the tiredness of a driver of a motor vehicle, comprising the steps of: detecting a number of physical magnitudes related to movement of the motor vehicle, determining from the physical magnitudes detected a number of fatigue-determining magnitudes among a set of possible fatigue-determining magnitudes, choosing a model for assessing a driver's tiredness out of a group of such models based on whichever among the set of fatigue-determining magnitudes it has been possible to determine, inputting the fatigue-determining magnitudes determined into the chosen model, using the model to calculate a fatigue value, comparing the fatigue value with at least one fatigue threshold, and indicating tiredness of the driver if the fatigue value exceeds the fatigue threshold.

According to a second aspect of the invention, this object is also achieved by a device for assessing the tiredness of a driver of a motor vehicle, comprising: a magnitude determination unit adapted to determining a number of fatigue- determining magnitudes among a set of possible fatigue-determining magnitudes, from a number of detected physical magnitudes related to movement of the motor vehicle, a model selector unit adapted to choosing a model for assessing a driver's tiredness among a group of such models based on whichever among the set of fatigue-determining magnitudes it has been possible to determine, inputting the fatigue-determining magnitudes determined into the chosen model, using the model to calculate a fatigue value, and

a comparison unit adapted to comparing the fatigue value with at least one fatigue threshold, and indicating fatigue of the driver if the fatigue value exceeds the fatigue threshold.

According to a third aspect of the present invention, this object is also achieved by a motor vehicle comprising a number of detectors adapted to detecting a number of physical magnitudes related to movement of the motor vehicle, a fatigue indication unit and a device for assessing the tiredness of a driver of the motor vehicle according to the second aspect.

According to a fourth aspect of the present invention, this object is also achieved by a computer programme product for assessing the fatigue of a driver of a motor vehicle, comprising a computer programme code for causing a computer, when loaded with the computer programme code, to: determine a number of fatigue-determining magnitudes among a set of possible fatigue-determining magnitudes, from a number of detected physical magnitudes related to movement of the motor vehicle, choose a model for assessing a driver's tiredness among a group of such models based on whichever among the set of fatigue-determining magnitudes it has been possible to determine, input the fatigue-determining magnitudes determined into the chosen model, use the model to calculate a fatigue value, compare the fatigue value with at least one fatigue threshold, and indicate tiredness if the fatigue value exceeds the fatigue threshold.

The invention has a number of advantages. It makes it possible to identify whether a driver is tired or not without having to monitor him/her, which would be both complicated and expensive. In addition, according to the invention, the model best suited to the occasion is chosen, which makes fatigue detection more reliable than would otherwise be the case.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention is described below in more detail with reference to the attached drawings, in which:

Fig. 1 depicts schematically a vehicle travelling on a road,

Fig. 2 depicts a block diagram of a fatigue detection device according to the invention,

Fig. 3 depicts a flowchart of a procedure for fatigue detection according to the invention, and

Fig. 4 depicts schematically a computer programme product in the form of a CD ROM disc containing a computer programme code for implementing the present invention.

DETAILED DESCRIPTION OF EMBODIMENTS

The present invention is adapted to assessing the tiredness of a driver of a motor vehicle on the basis of a number of physical magnitudes of the motor vehicle which are detected. In many cases it is important to assess whether the driver of the motor vehicle is tired. This may be important inter alia for traffic safety reasons.

Fig. 1 depicts a motor vehicle 10. The motor vehicle is preferably a truck, but it should be noted that the present invention is not limited to trucks but usable in all motor vehicles which are controlled by a driver. In Fig. 1 the motor vehicle 10 is being driven on a road 12. Also on the road are a central line 14 and side lines 15. The vehicle is equipped with a first detector 16 comprising a camera which monitors the central line 14 and at least one of the side lines 15. This camera is used here for detecting a first physical magnitude which is the vehicle's lateral position. It should be noted that the vehicle has a plurality of sensors for detecting a number of different physical magnitudes.

Fig. 2 depicts a block diagram of a number of different units provided in the vehicle 10 for determining whether the driver is tired or not. There are a number of detectors as follows: a first detector 16 detects the vehicle's lateral position, a second detector 18

detects the vehicle's lateral velocity, a third detector 20 detects the vehicle's lateral acceleration, a fourth detector 22 is connected to the vehicle's steering wheel and detects the steering wheel angle, a fifth detector 24, also connected to the steering wheel, detects the angular velocity of the steering wheel, a sixth detector 26 detects the steering wheel torque, a seventh detector 28 detects the vehicle's yaw angle and an eighth detector 30 detects the vehicle's speed. A first group of these detectors, viz. the first to third detectors 16 - 20, are adapted to detecting physical magnitudes related to the vehicle's lateral movement, while a second group of detectors, viz. the fourth to sixth detectors 22 - 26, are adapted to detecting physical magnitudes related to movements of the steering wheel. It should be noted that the invention is not limited to these detectors and that more or fewer may be used according to the invention. It is also possible to detect other physical magnitudes, e.g. the force exerted on the steering wheel and corresponding values, force, acceleration etc. exerted on the vehicle axle, and also acceleration in the direction of travel, brakes etc.

All these detectors 16 - 30 are here connected to a device 32 for assessing the tiredness of a driver of the vehicle 10, and more specifically to a magnitude determination unit 34 in that device 32. This magnitude determination unit 34 is itself connected to a model selector unit 38. The model selector unit 38 is connected to a model memory 36 which contains a group of fatigue determination models, and to a comparison unit 40. Finally, the comparison unit 40 is connected to a fatigue indication unit 42. This fatigue indication unit 42 is not part of the device 32 here, but it should be noted that it might very well be part of the latter. The fatigue indication unit 42 may be implemented in the form of a display and/or by means of a loudspeaker for letting a driver of the motor vehicle 10 know that he/she is tired.

The present invention is explained below in more detail with reference also to Fig. 3, which depicts a flowchart of a number of steps performed in a procedure according to the present invention.

The procedure begins with a driver sitting in the vehicle 10 and starting to drive the vehicle. When this happens, the detectors 16 - 30 detect the various physical

magnitudes described above, step 44. The magnitudes detected are then passed on to the magnitude determination unit 34. From these physical magnitudes detected the magnitude determination unit 34 thereafter determines a number of fatigue- determining magnitudes, step 46. In the example here described, these fatigue- determining magnitudes are as follows:

LATVAR: variance and standard deviation of lateral position

MEANPOS: mean lateral position

PATHDEV: area of deviation, representing a relationship between lateral position and mean lateral position and calculated as the area between these values over time

TLC: time to crossing the side line

LANEX: number of side and central line crossings

ELLIPSE: represents the steering wheel angle plotted against the steering wheel angular velocity in a plane. These two detected magnitudes are plotted with respect to their mean values and this fatigue-determining magnitude is determined by the distance from the origin in that plane

NEW STEXED: proportion of time when the steering wheel angular velocity exceeds a given first threshold value

MY S TEXED: a mean value of the steering wheel angular velocity over a given period of time

WW_NMRHOLD: proportion of time when the steering wheel angle θ is kept steady at any desired angle

MY-NMRHOLD: proportion of time when the steering wheel angle θ is kept steady about zero degrees

SDEV: standard deviation for the steering wheel angle

STVELV: variance of the steering wheel angular velocity

SWDR: number of times the steering wheel changes direction over a certain period of time

AMP_D2_θ: area between the steering wheel angle θ and the steering wheel angle mean value multiplied by the time during which the steering wheel angle is on the same side of the steering wheel angle mean value.

REACTIM: reaction time calculated as the time difference between the lateral acceleration and a subsequent steering wheel torque

DEGOINT: degree of coupling, representing the relationship between steering wheel torque and lateral acceleration and calculated as the area between these values over time.

All these fatigue-determining magnitudes are among a set of fatigue-determining magnitudes which are partly used for choosing which out of a group of fatigue determination models should be used, and they also form part of the various fatigue determination models.

There is also a further fatigue-determining magnitude which is not part of that set but is still calculated, namely:

FREQANA: frequency analysis in a certain frequency band of at least one detected magnitude. This fatigue-determining magnitude is determined by Fourier transformation of the detected physical magnitude. It is possible here to do this for a number of different detected magnitudes such as lateral acceleration, lateral position, speed, steering wheel angle, steering wheel torque and yaw angle, in order to determine the energy in them.

It should also be noted that it is not always possible to determine all of the magnitudes among the set of fatigue-determining magnitudes, particularly those relating to crossing the central and side lines.

When the fatigue-determining magnitudes have been determined, including the further fatigue-determining magnitude FREQANA, the further fatigue-determining magnitude FREQANA is compared, in the magnitude determination unit 34, with a first threshold value, i.e. the energy in a certain frequency band is compared with a threshold energy. If the energy of the further fatigue-determining magnitude is higher than this first threshold, step 48, tiredness is indicated, step 60, and the fatigue determination unit emits a fatigue indication signal to the fatigue indication unit 42 to let the driver know

that he/she is tired. If the energy is not higher than this first threshold, step 48, the values of the fatigue-determining magnitudes determined are sent, as well as the further fatigue-determining magnitude FREQANA, to the model selector unit 38. The model selector unit 38 then looks at the fatigue-determining magnitudes available among the set and, depending on which it has thus been possible to determine, a fatigue determination model is chosen, step 50. There may here be as many models as there are possible simultaneous combinations of fatigue-determining magnitudes. Some of the fatigue-determining magnitudes are available all the time, but two in particular may only be determined sometimes. These two in the above example are LANEX and TLC. If neither of these is available a first model is chosen, if only one of them is available a second model is chosen, if only the second of them is available a third model is chosen, and if both are available a fourth model is chosen. Another fatigue-determining magnitude which may in some situations not be available is ELLIPSE. This might mean that there are four further model combinations, but in the example here described these are reduced to only two further models, since two of these further models in this example do not meet the requirement that the number of data points has to be greater than the number of independent fatigue-determining magnitudes.

The models chosen have the general formula:

Y - βo + βiXi + βiXl + . - . + βn Xn + ε

where Y is a threshold value, βo is a constant, βi . β _n are coefficients, x are the fatigue- determining magnitudes and ε is the stochastic error.

All the models are constructed according to this principle but have different coefficients and constants, and each model may also contain different fatigue- determining magnitudes. The models are therefore static, whereas the fatigue- determining magnitudes are determined continuously in real time.

Once a model has been chosen from the group, the model selector unit 38 takes the chosen model from the memory 36 and inputs the relevant fatigue-determining magnitudes into the chosen model, step 52. The threshold value Y is then calculated according to the model, step 54, and is thereafter transferred to the comparison unit 40. This unit 40 compares the threshold value Y with a second fatigue threshold, step 56, and if the value is over the fatigue threshold, step 58, tiredness is indicated, step 60, and the magnitude determination unit emits a fatigue indication signal to the fatigue indication unit 42 to let the driver know that he/she is tired. If the value is not above the threshold, nothing happens, step 62. This procedure is thereafter repeated at regular intervals, e.g. every ten minutes, to determine tiredness of the driver.

The present invention has a number of advantages. It makes it possible to identify whether a driver is tired or not without having to monitor him/her, which would be both complicated and expensive. In addition, according to the invention, the model best suited to the occasion is chosen, which makes fatigue detection more reliable than would otherwise be the case.

The following is an example of the configuration of the various models in an embodiment:

Y ₀ = 7.79 -2.4*SWDR + 10387.71 *PATHDEV + 0.64*LATVAR

Y ₁ = -17.55 - 1.57*SWDR + 1.04*LATVAR + 24.86*MEANPOS

Y ₂ = 2.45 - 4.29*WW_NMRHOLD + 470594.05* DEGOINT + 0.71*ELLIPSE +

7.69*REACTIM - 3.83*SDEV

Y ₃ = 0.71 - 0.57*STVELV + 1.40*LATVAR - 2.58*SDEV + 7.69*REACTIM

Y ₄ = 8.09 - 4.72*WW_NMRHOLD + 9936.37*PATHDEV

Y ₅ = 6.49 + 211889.40*LANEX + 0.43*LATVAR

The magnitude determination unit, the model selector unit and the comparison unit in the device for assessing tiredness of a driver preferably take the form of a processor with one or more programme memories containing a computer programme code which executes the procedure according to the invention. This may be done by a vehicle

computer. The programme code may also be provided in the form of a computer programme product, which may be in the form of a portable memory device such as CD ROM disc. Such a disc 64 is depicted schematically in Fig. 4. The programme code may also be provided in the form of a pure computer programme code, which may be provided on a server and be downloaded therefrom to a vehicle. A device according to the invention is implemented when such a programme code is loaded into a vehicle computer either from a server or from a portable memory device. It should be noted, however, that the invention is also implementable by a suitable combination of logic circuits. Finally, the model memory may be arranged, for example, as an RAM or an ROM memory.

It should be noted that some of the above fatigue-determining magnitudes initially determined may be stored in a memory, e.g. the memory 36. They may subsequently be used for adjusting such fatigue-determining magnitudes determined later, i.e. magnitudes of the same type. The following are examples of such fatigue-determining magnitudes:

LATVAR

PATHDEV

NEW STEXED

MY STEXED

WW NMRHOLD

MY-NMRHOLD

SDEV

AMP_D2_ θ

REACTIM

DEGOINT

By the aforesaid adjustment it is possible to take into account the individual driver's driving technique, thereby affording the possibility of more reliable fatigue detection.

The models used were arrived at by regression analysis. This involved the various magnitudes being determined in a simulator while at the same time test persons used the simulator to simulate driving. These persons thereafter indicated their tiredness on a scale. Based on those indications, the models were then arrived at by regression analysis. The forward choice principle was thereafter applied to determine which magnitudes should form part of the models. The magnitudes to be included did not need to be correlated with one another nor exhibit any internal relationship. Thereafter the models which best matched the test persons' indications were chosen and were optimised, which was done by the least-squares method. The models which matched best were then subjected to shaving of peaks which correlated with and exceeded the peaks of the test persons' indications. That is how the models were arrived at and optimised. The resulting models thus constituted a group of models and were then stored in the memory 36. The models are thus static and the same for every vehicle.

As indicated in the paragraph above, a model need therefore not use all of the fatigue- determining magnitudes. In an embodiment of the invention, all that is determined is therefore the fatigue-determining magnitudes used in the various models which may be chosen and are used for choosing models.

It should also be noted that the present invention may be applied without using the further fatigue-determining magnitude.

The above example uses only one threshold value for the respective first and second thresholds for indicating tiredness. It should be noted, however, that it is possible to use a plurality of different threshold levels in relation to the first and second threshold comparisons in order to indicate different degrees of tiredness. It is possible here to indicate very fresh, fresh, neither fresh nor tired, rather tired and very tired. It is also possible for these various degrees to be used in different ways for letting the driver know how tired he/she is, e.g. by using different combinations of sound and image and different sound volume levels. It should also be noted that additional further fatigue- determining magnitudes may be determined by frequency analysis of a corresponding detected physical magnitude and be compared with corresponding thresholds. It is

also possible to send a fatigue indication from the vehicle to a server, e.g. a server belonging to a haulage company.

As may be seen from what is stated above, the invention may be varied in all kinds of ways. The present invention is therefore only to be limited by the following claims.