GB2504586A - A vehicle having a system and method of scheduling driver interface tasks - Google Patents

Abstract

A vehicle's dynamic handling state, driver inputs to the vehicle, etc. may be examined to determine one or more measures of driver workload. Driver interface tasks may then be delayed and/or prevented from executing based on the driver workload so as to not increase the driver workload. Alternatively, driver interface tasks may be schedule for execution based on the driver workload and caused to execute according to the schedule, for example, to minimize the impact the executing driver interface tasks have on driver workload. For example, driver interface tasks may include forwarding in-coming calls to voice mail and/or generating an audio output and/or generating a visual output and/or generating a tactile output. Some of the interface tasks are delayed or prevented from being executed based on a maximum value of generated parameters and/or if the driver workload fall within a predetermined range of values.

Description

SYSTEMS AND METHODS FOR SCHEDULING DRIVER

INTERFACE TASKS BASED ON DRiVER WORKLOAD

BACKGROUND

Certain vehicles may provide infotainment information, navigation information, etc. to enhance the driving experience. As the interaction between dnvers and these vehicles increases, it maybe beneficial to facilitate such interaction without increasing driver workload.

SUMMARY

Measures of driver workload maybe determined from vehicle, driver andlor environmental information. Certain driver interface tasks may be selectively delayed or prevented from executing based on the determined driver workload. Alternatively, driver interface tasks may be scheduled for execution based on the determined driver workload and then caused to execute according to the schedule.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure I is a block diagram of an example hybrid workload estimation system.

Figure 2 is a plot of example vehicle speed, traction and braking profiles.

Figures 3A through 3C are plots of example vehicle motion states of yaw rate and sideslip angle.

Figures 4A through 4C are plots of example yawing. longitudinal, and side slipping handling limit margins.

Figure 5 is a plot of example vehicle speed, traction and braking profiles.

Figures ÔA through ÔC are plots of example vehicle motion states of yaw rate and sideslip angle.

3 0 Figures 7A through 7C are plots of example yawing, longitudinal, and side slipping handling limit margins.

Figures 8 and 9 are plots of example final handling limit margins and risk.

Figures 10 and II are plots of example accelerator pedal position for high demand circumstances and low demand circumstances respectively.

Figures 12 and 13 are histograms of the standard deviation of the accelerator pedal position of Figures 10 and ii respectively.

Figure 14 is a plot of curves fit to the histograms of Figures 12 and 13.

Figures iSA through 15D are example plots of accelerator peda' position, steering wheel angie, Driver Control Action (DCA) Index, and vehicle speed respectively.

Figures l6A through l6C are example plots of turn indicator activity, air conditioning control activity, and Instrument Panel (IP) Index respectively.

Figure 17 is a schematic diagram of a vehicle following another.

Figures 18, 19 and 20 are example plots of vehicle speed, closing velocity and range, respectively.

Figures 21 and 22 are example plots of headway and Headway(HW) Index respectively.

Figures 23A through 23E are example plots of a Rule-Based Index, IP Index, DCA Index, composite Workload Estimation (WLE) hidex, and vehicle speed respectively.

Figure 24 is an example plot of membership functions for characterizing driver demand based on the WLE Index.

DETAILED DESCRIPTION

1. Introduction

Driver workload/demand may refer to the visual, physical and cognitive demand that secondary activities such as infotainment, phone, proactive recommendations. etc. place on the driver above and beyond the primary activity of driving.

Drivers may sometimes incorrectly assume that they are able to divide their attention between the primary activity of driving and the secondary activities discussed above. Estimating the driving demand, therefore, may be of particular value if used to modulate communication and vehicle system interactions with the driver. Complex driving contexts, however, may require innovative prognostic approaches to driver workload estimation. Development of intelligent systems that enable the identification of driver workload may be useful in tailoring human machine 3 0 interface (HMI) outputs to the driver.

For continuous workload estimation, it may be useful to design an estimator that predicts the workload under different driving contexts andlor drivers. Adaptation of vehicle cabin communication services may be based on the context within which the driving demand is predicted and the value of the services to the driver. In addition, characterizing the driver workload over periods of time (e.g., long term characterization) may be beneficial. Such assessment of the driver workload may allow vehicle cabin communication technologies to not only be suppressed or delayed during high workload periods, but in addition tailored to the long driving demand.

Certain embodiments described herein may provide methods and systems for Workload Estimation (WLE). The WLE may perform state estimation/classification of the driver workload from observable vehicle, dnver and environment data for adaptive real-time HMI task management. The WLE, in certain situations, may use separate real-time techniques and/or employ a real-time hybrid approach to workload estimation. A rule-based algorithm, for example. may be supplemented with additional continuous prediction of dryer workload based on the driver, vehicle 1 0 and environment interactions. The WLE algorithms may incorporate specialized learning and computational intelligence techniques to compute and predict an aggregated WLE Index (e.g., a continuous signal representing a workload load estimate for the driver). The driver's driving demand may be inferred, in certain instances, from observable vehicle information including variations in speed, acceleration, braking, steering, headway, instrument panel andlor center stack interaction, etc. The WLE Index may be used, for example, to set/avoid/limit/tailor voice commands andlor other tasks/information presented to the driver to improve functionality. Certain information for the driver may be limited/tailored/blocked during demanding vehicle handling manoeuvres, in hazardous driving environments, during periods of high activity with the instrument panel, etc. Intelligent hybrid algorithmic approaches may account for long-term as weB as short-term driver actions. WLE hybrid methods may capture the dnver events, situations and behaviour for coordinating vehicle to driver communications. These and other techniques described herein may assist in predicting driver increasing/decreasing cognitive conditional states and use existing vehicle sensors.

The WLE Index may also allow a hierarchy of communication to be presented to the driver based on the driving demandl workload. Message priority (e.g., low, high, etc.) may determine whether the message is delivered to the driver during a particular instant based on the predicted load. Specific HMI information may also be presented to the driver based on the driver's long-term driving demand. Alternatively, a Hybrid WLE framework may incorporate UPS and 3 0 digital map databases to account for road scenario situations and conditions. Information about the driver's physiological state including heart-rate, eye-gaze, and respiration may, in addition, be incorporated as inputs to the WLE framework for anomaly detection. In other examples. the predicted WLE index may be communicated to the dnver to remind them to avoid secondary tasks under high workload conditions. Other scenarios are also possible.

Figure 1 is a block diagram of an embodiment of a WLE system 10 for a vehicle 11.

The system 10 may include a rule-based workload index sub-system 12, a vehicle, driver and environment tracking and computation workload index sub-system 13, a context dependent workload index aggregation sub-system 14, and an overall aggregation/WLE long term characterization sub-system 16. The sub-systems 12, 13, 14, 16(individuaflyor in combination) may be implemented as one or more controllers/processing devices/etc.

The sub-system 12 (as explained in section VU below) may take as input vehicle information, driver information and/or environmental information (available, for examp'e, from the vehicle's controller area network (CAN)), and output a rule-based index representing dryer workload. The sub-system 13 (as explained in sections III through VI below) may take as input vehicle information, driver information, and/or environmental information (available, for example, from the vehide's CAN), and output one or more contirnious indices (e.g., Handling Limit (ML) Index, Driver Control Action (DCA) Index, Instrument Panel (IP) Index, Headway (HW) Index) representing driver workload. The sub-system 14 (as explained in section VIII below) may take as input the index/indices generated by the sub-system 13, and output a Tracking (T) Index. The sub-system 16 (as explained in section VIII below) may take as input the rule-based index and T index, and output a WLE Index and/or (as explained in section IX below) a long term characterization of the WLE Index.

The system 10, in other embodiments, may lack the sub-systems 12, 14, 16. That is, certain embodiments may be configured to only generate one or more workload indices. As an example, the system 10 may be configured to only generate the IP Index based on certain vehicle information (discussed below). No aggregation is necessary in these circumstances as there is only a single measure of driver workload. Hence the WLE Index, in this example, is the IP index. The dispatcher 18, in these and other embodiments. may be configured to generate the long term characterization of the W/LE Index. Other arrangements are also possible.

The WLE Index may be sent to a dispatcher 18, which may be implemented as one or more controllers/processing devices/etc. The dispatcher 18 (as explained in section X below) may act as a filter-preventing/delaying information to be communicated to the driver from reaching the driver based on the WLE Index. For example, if the WLE Index is greater than 0.8, all information intended for the driver may be blocked. If the WLE Index is approximately 0.5, only entertainment-type information may be blocked, etc. The dispatcher 18 may also schedule delivery of information to be communicated to the driver based on the WLE Index. For example. vehicle maintenance information, text-to-speech readouts, incoming calls, etc. may be delayed during periods of high workload. In addition, the dispatcher 18 may enable vehicle outputs to be tailored to the driver based on a long term WLE Index characterization as discussed in more detail below.

For example, the output of certain vehicle systems including cruise control, adaptive cruise control, music suggestion, configurable HMI, etc. maybe based on the long term driving demand.

The driver's workload state may be inferred from observable vehicle information including variations in speed, acceleration, braking, steering, headway, instrument panel interaction, etc. Table 1 lists example featureslmetrics related to driver workload load.

Table 1

Example Features/Metrics Related to Driver Workload

__________________________________ _________________________________________

Metric Behavioral Effect Intended to Quantify Mean Speed Large speed increase/reduction Maximum Speed Large speed increase Mean Time Headway (Gap Time) Reduced headway Minimum Time Headway Reduced minimum headway Brake Reaction Time (BRT) Reduced BRT Brake Jerks Jiwreased frequency Steering Wheel Reversal Rate Increased frequency of small reversals interaction with 1P (e.g., pressing Increased frequency of IP buttons) Traffic Density Increased density Driving Location New driving environment Mean Speed Large speed increase/reduction Maximum Speed Large speed increase Tables 2a and 2b list example information which may be available/accessible via CAN as known in the art. The following information may be used as inputs to any of the algorithms described herein.

Table 2a

Example Information Available via CAN Accelerator pedal position Microphone inputs Steerng wheel angle Cup holder sensor Seat sensor Vehicle speed Turn signal Yaw rate Defrost signal Lateral acceleration Temperature control LongitiLdinal acceleration Headlamp status Wheel speeds High beam status Throttle position Fog lamp status Master cylinder pressure Radio tuner command Driver request torque Wiper status Bus axle torque Gear position Bus torque distribution Rain sensor Roll rate Configurable HMI Sideslip angle Touch HMI Relative roll angle

Table 2b

Example System Information Accessible via CAN Traction Control System Anti-Lock Braking System Electronic Stability Control Adaptive Cruise Control Collision Mitigation by Braking Blind Spot Monitoring Automatic Parking Aid II. A Brief Discussion Of Vehicle Stability Controls A vehicle's handling determines the vehicle's ability to corner and maneuver. The vehicle needs to stick to the road with its four tire contact patches in order to maximize its handling capability. A tire which exceeds its limit of adhesion is either spinning, skidding or sliding. A condition where one or more tires exceed their limits of adhesion may be called a limit handling condition and the adhesion limit may be called a handling limit. Once a tire reaches its handling limit, the average driver is usually no longer in control. In the so-called understeer case, the car under performs a driver's steering input, its front tires pass their handling limit, and the vehicle continues going straight regardless of the driver's steering request. In the so-called oversteer case, the car over performs the driver's steering inputs, its rear tires pass their handling limit, and the vehicle continues spinning. For safety purposes, most vehicles are built to understeer at their handling limits.

In order to compensate vehicle control in case a driver is unable to control the vehicle at or beyond the handling limit, electronic stability control IESC) systems are designed to 1 0 redistribute tire forces to generate a moment that can effectively turn the vehicle consistent with the driver's steering request. Namely, to control the vehicle to avoid understeer and oversteer conditions.

Since its debut in 1995, ESC systems have been implemented in various platforms.

Phasing in during model year 2010 and achieving full installation by model year 2012, Federal Motor Vehicle Safety Standard 126 requires ESC systems on any vehicle with a gross weight rating below 10,000 lb. ESC systems may be implemented as all extension of anti-lock braking systems (ABS) and all-speed traction control systems (TCS). They may provide the yaw and lateral stability assist to the vehicle dynamics centered around the driver's intent. It may also proportion brake pressure (above or below the driver applied pressure) to individual wheel(s) so as to generate an active moment to counteract the unexpected yaw and lateral sliding motions of the vehide. This leads to enhanced steering control at the handling limits for any traction surface during braking, accelerating or coasting. More specifically, current ESC systems compare the driver's intended path to the actual vehicle response which is infelTed from onboard sensors. If the vehicle's response is different from the intended path (either understeenng or oversteering), the ESC controller applies braking at selected wheel(s) and reduces engine torque if required to keep the vehicle on the intended path and to minimize loss of control of the vehicle.

A limit handling condition can be detected using data already existing in ESC systems, so new sensors may not be required. Consider, for example, a vehicle equipped with an ESC system using a yaw rate sensor, a steering wheel sensor, a lateral accelerometer, a wheel speed 3 0 sensors, a master cylinder brake pressure sensor, a longitudinal accelerometer, etc. The vehicle motion variables are defined in the coordinate systems as defined in ISO-8855, where a frame fixed on the vehide body has its ver ical axis up. its axis along the longitudinal direction of the vehicle body, and its lateral axis pointed from the passenger side to the driver side.

S

Generally speaking, vehicle level feedback controls can be computed from individual motion variables such as yaw rate, sideslip angle, or their combination together with arbitrations among other control commands such as driver braking, engine torque request, ABS and TCS. Vehicle level control commands are discussed in the following.

The well-known bicycle model captures the vehicle dynamics, its yaw rate co along the vertical axis of the vehicle body and its sideslip angle fi, defined at its rear axle, and obeys the following equations: I,th =bfCf(fir +bwv;' 5)+brCrflr +M, (1) M(1rflr +Vrflr +bth, +w,vj= cf(fl7 +bQv' -8)-c,fl, where v1 is the vehicle's travel speed, M and 4 are the total mass and the yaw moment of inertia of the vehicle, Cf and c,-are the cornering stiffness of the front and rear tires, bf and b. are the distances from the center of gravity of the vehicle to the front and rear axles, h = hj÷ hr. M is the active moment applied to the vehicle, and ô is the front wheel steering angle.

A target yaw rate co, and a target sidesl ip angle f3, used to reflect the driver's steering intent can be calculated from (1) using the measured steering wheel angle ö and the estimated travel velocity v as the inputs. hi such a computation, we assume that the vehicle is driven on a road of normal surface condition (e.g.. high friction level with nominal cornering stiffness cf and c,). Signal conditioning, filtering and nonlinear colTections for steady state limit 2 0 cornering may also be performed in order to fine tune the target yaw rate and the target sideslip angle. These calculated target values characterize the driver's intended path on a normal road surface.

The yaw rate feedback controller is essentially a feedback controller computed from the yaw error (the difference between the measured yaw rate and the target yaw rate). If the vehicle is turning left and o? w + Ozdbo. (where 0zJbo. is a time varying deadb and), or the vehicle is turning right and w - the vehicle is oversteering and activating the oversteer control function in ESC. For instance, the active torque request (applied to the vehicle for reducing the oversteer tendency) might be computed as in the following during a left turn: M, = min(O,-k0jw -w wZdbJ) (2) during a right turn: M = max(O,-k0 (w -+ (D,dhOi) where Ic, is a speed dependent gain which might be defined as in the following = + (v --(3)

-

with parameters k0, kiN, kd,,,,, Vlrdh?, Vxdhu being tunable.

OJ -0Zdbus (where cozdbus is a time varying deadband) when the vehicle is turning left or cv.? cv. + w,jh,,, when the vehicle is turning right, the understeer control function in ESC is activated. The active torque request can be computed as in the following during a left turn: M, = max(O,-k,,, (oi -+ (4) during a right turn: M = min(O,-k, (w --where k,,, is a tunahie parameter.

The sideslip angle controller is a supplementary feedback controller to the aforementioned oversteer yaw feedback controller. It compares the sideslip angle estimation /3. to the target sideslip angle /3,-,. If the difference exceeds a threshold fidh. the sideslip angie feedback 2 0 control is activated. For instance, the active torque request is calculated as in the following during a left turn, fir »= 0: A1_ = nnn(o. k, (fir -B -Brdh) -ksscmpflrcmp) (5) during a right turn, fir <0: M, = rnax(0, k5, (fi -B,., + B,db) -where 1c and k,scmp are tunable parameters and /3,*cm,, is a compensated time derivative of the sideslip angle.

Other feedback control terms based on variables such as yaw acceleration and sideslip gradient can be similarly generated. When the dominant vehicle motion variable is either the yaw rate or the sideslip angle, the aforementioned active torque can be directly used to determine the necessary control wheel(s) and the amount of brake pressures to be sent to corresponding control wheel(s). If the vehicle dynamics are dominated by multiple motion variables, control arbitration and prioritization will be conducted. The final arbitrated active torque is then used to detemñne the final control wheel(s) and the colTesponding brake pressure(s). For example, during an oversteer event, the front outside wheel is selected as the control wheel, while during an understeer event, the rear inside wheel is selected as the control wheel. During a large side-slipping case, the front outside wheel is always selected as the control wheel. When both the side slipping and oversteer yawing happen simultaneously, the amount of brake pressure may be computed by integrating both yaw error and the sideslip angle control commands.

Besides the above cases where the handling limit is exceeded due to the driver's steering manoeuvres, a vehicle can reach its limit handling condition in its longitudinal motion direction. For example, braking on a snowy and icy road can lead to locked wheels, which increases the stopping distance of the vehicle. Open throttfing on a similar road can cause the drive wheels to spin without moving the vehicle forward. For this reason, the handling limit may also be used for these non-steering driving conditions. That is, the conditions where the tire longitudinal braking or driving forces reach their peak values may also be included in a definition of the handling limit.

The ABS function monitors the rotational motion of the individual wheels in relation to the vehicle's travel velocity, which can be characterized by the longitudinal slip ratios 2, with i = 1, 2,3,4 for the front-left, front-right, rear-left arid rear-right wheels, computed as in the following 1= -1 max((v + Wztj)cos(8)+ (v + wb1)sin(8),v1) 2, = K,W. -l (6) max((v + Qif)cos(ö) + + oLb)sin(8), ___________ K4w1 -i max(vr -W;!r v, ) max(v + mar' Vrrfln) where tf and tr are the half tracks for the front and rear axles, w1 is the wheel speed sensor output, #q is the 1h wheel speed scaling factor, v is the lateral velocity of the vehicle at its c.g. location, and is a preset parameter reflecting the allowable minimum longitudinal speed. Notice that (6) is only valid when the vehicle is not in the reverse driving mode. When the driver initiated braking generates too much slip (e.g., -2»= ILL,j, = 20%) at a wheel, the ABS moduth will rdease the brake pressure at that wheel. Similarly, during a large throttle application causing a large slip on the i" driven wheel, the TCS module will request engine torque reduction and/or brake pressure applied to the opposite wheel at the same axle. Consequently, ABS or TCS activations can be predicted by monitoring how close 2s are from 2k" and),.

111. Handling Limit Index While the aforementioned LSC (including ABS and IfS) is effective in achieving its safety goal, further enhancement is still possible. For example, augmentation of ESC systems may be desirable for roll stability control. The appropriate colTection which ESC tries to make, however, maybe counteracted by the driver or amhient conditions. A speeding vehicle whose tire forces go far beyond the traction capability of the road/tires might not be able to avoid an 1 0 understeer accident even with ESC intervention.

Generally speaking, accurate determination of the handling limit conditions would typically involve direct measurement of road and tire characteristics or intensive information from many related variables if direct measurements are not available. Currently, both of these methods are not mature enough for real-time implementation.

Due to their feedback feature, ESC systems may he configured to determine the potential limit handling conditions through monitoring the motion variables (vehicle handling parameters) of a vehicle such as those descrihed in the last section. When the motion variahles deviate from their reference values by a certain amount (e.g., beyond certain deadbands). the ESC systems may start to compute differential braking control command(s) and delermine control wheel(s). ftc corresponding brake pressure(s) is then sent to the control whee'(s) to stabilize the vehicle. The starting point of the ESC activation can be thought of as the beginning of the handling limit.

More specifically, we may define a relative handling limit margin h, as in the following h= rfx«=x<O (7) C) othenvise where xis the deviation of a motion variable from its reference value and [x, k defines the deadhand interval within which x falls without initiating the ESC, ABS or TCS. x can be any of the control variables defined in the ast section (or any other suitaNe control variable).

The benefit of h1 defined in 7) is that the driving condition can be quantitatively characterized into different categores. For instance, when h 10%, the driving condition may be categorized as a red zone condition where the driver needs to have special attention or take some special actions (e.g., slowing down the vehicle); when 10% <h< 40%, the driving condition may be categorized as a yellow zone condition which needs some level of special attention from the driver; when 40% c h. 100%, the driving condition maybe characterized as a normal condition.

1 0 In the normal condition, the driver needs only to maintain his normal driving attention. Of course, other ranges may also he used.

More specifically, let us use the control variables computed in the last section to discuss the computation of hs. The vehicle's yaw handling limit margin during oversteer situations h05 (where w> w, when the vehicle is turning to the left and w> wv when the vehicle is turning to the right) can be computed from (7) by setting x = -w, and = = -x, where Wzdbos is the oversteer yaw rate deadhand as defined in (2).

Similarly, the vehicle's yaw handling limit h,15 for understeer situations can he computed from (7) by setting x = co -coY, and = = -x where 0zjbu. is the understeer yaw rate deadhand as deuined in (4). Notice that the aforementioned deadhands might he lunctions of the vehide speed, the magnitiLde of the target yaw rate, the magnitude ol the measured yaw rate, etc. The deadhands for the understeer situation (x c 0) and the oversteer situation (x > 0) are different and they are tunable parameters.

The vehicle's sideslip handling limit margin hs51Q% can be computed from (7) by setting x = fir - and i = flr1m = -x.

The longitudinal handling limits of the vehicle involve the conditions when either the driving or braking force of the tires approaches the handling limit. The traction control handling limit margin for the i'' driven wheel h?(s can be computed from (7) by setting x = x = 0, and k = 2rn The ABS handling limit margin for the wheel h can also he computed from (7) by setting x = 2e, = 2bp' and k = 0. l'he final traction and braking handling limit margins maybe defined as = mm hABS,h5 = mm hT(s (8) e{1,23,4} Notice that further screening conditions may be used in computing the aforementioned handling limit margins. For instance, one of the following or the combination of some of the following conditions might be used to set the handling limit margin as 0: a magnitude of the target yaw rate is beyond a certain threshold; a magnitude of the measured yaw rate is greater than a certain threshold; a driver's steering input exceeds a certain threshold; or, extreme conditions such as the vehicle's cornering acceleration is greater than O.Sg, the vehicle's deceleration is greater than 0.7g. the vehicle is driven at a speed beyond a threshold (e.g., 100 mph), etc. In order to test the aforementioned handling limit margin computations and verify 1 0 their effectiveness with respect to known driving conditions, a vehicle equipped with a research ESC system developed at Ford Motor Company was used to conduct vehicle testing.

For the driving condition profiled by the vehicle speed. throttling, and braking depicted in Figure 2, the measured and computed vehicle motion vanablcs are shown in Figures 3A through 3C. The corresponding individual handling limit margins huc. h05. hTcc. hAns, and /is are shown in Figures 4A through 4c. This test was conducted as a free foim slalom on a snow pad with all ESC computations running. [he brake pressure apply was turned off in order for the vehicle to approach the true limit handling condition.

For another test, the vehicle was dnven on a road surface with high friction level.

The vehicle speed, traction and braking profiles for this test are depicted in Figure 5. The vehicle motion states arc shown in Figures 6A through 6C. l'he corresponding individual handling limit margins h05. hTcs, h411,, and kSSRA are shown in Figures 7A and 7B.

An envelope variable of all the individual handling limit margins is defined as = niin{h0.huc,hT(. ,h48,hscRA} (9) Considering that sudden changes in thc envclopc handling limit margin might be due to signal noises, a low-pass filter F(z) is used to smooth so as to obtain the final handling Limit (IlL) Index or margin hJ'(z)h, (10) For the vehicle test data shown in Figure 2 and Figures 3A through 3C, the final handling limit margin is depicted in Figure 8, while for the vehicle test data shown in Figure 5 and Figures ÔA through 6C. the final handling limit mgin is depicted in Figure 9.

the HL Index may provide a continuous variable between 0 and 1 and indicate how close the driver is to the vehicle's handling limit (where a value of I indicates that the driver is at the vehicle's handling limit). This model-based IlL Index may provide particularly important driving demand infoimation during, for example. 1ow-road driving conditions.

Assuming that more visual. physica' and cognitive attention is required to maintain vehicle control as the vehicle approaches its handling limit, driver workload information may be 1 0 inferred from the I-IL hidex. As the workload on the driver increases, the HL hidex increases. As the workload on the driver decreases, the IlL Index decreases.

IV. Driver Control Action Index The Driver Control Action (DCA) Index may provide a continuous variable between 0 and 1 that indicates the total variability of the driver's control action with respect to, for example, acceleration, braking and steenng. Increased variability from the dnver's operational norm may reflect increased driving demand and vise-versa. The I)CA Index may this provide a measure of the variability (driving demand) associated with different drivers having different norms of vehicle control action.

Consider, for example, the impact of the accelerator pedal variahility on driving demand. RefelTing to Figures 10 and 11, real-time accelerator pedal positions are plotted versus time for example high demand and low demand circumstances, respectively. Considerably more variability of the accelerator pedal is evident in the high demand case relative to the low demand case.

The standard deviation of the accelerator pedal positions of Figures 10 and 11 are plotted respectively in Figures 12 and 13.

RefelTing to Figure 14, probabilistic fits to the distributions of Figures 12 and 13 are generated using a gamma function of the standard form y = f(xI a,h)= (11) where a is the scale factor and b is the shape factor. The dashed line represents the low driving demand standard deviation distribution and the solid line represents the high driving demand standard deviation distribution. These probabilistic distributions of the accelerator pedal variability show levels of distinction between the driving demand categories and present opportunities for classification. lor example. a standard deviation of 2% would have a higher probability of being characteristic of low driving demand, whereas a standard deviation of 10% would have a higher probability of being characteristic of high driving demand, etc. This technique may similarly be applied to brake pedal position, steering wheel angle. andlor other driver control action parameters.

hence, the DCA Index may estimate the driving demand based on the variability of the driver 1 0 action on the accelerator pedal, brake pedal, steering wheel, etc. The averages of the standard deviation variability shown in Figure 14 can change with different drivers. the DCA Index computation may account for these changing averages and compute the relative variability. The derivative of the driver inputs may also be incorporated to capture anticipatory action. This variance computation may be obtained fl-cm aaJysis of the determinant of the covariance for each of the factors (e.g., accelerator pedal position/rate, brake pedal position/rate, steering wheel angle position/rate, etc.) 1'he DCA Index, in certain embodiments, is computed by recursively ca'culating the determinant of the covariance affecting driving demand for each of the factors based on the following equations: Axk = -Ak (12) Xk+l =UüXk -I-axk (13) G r(1-.).Gk1 (14) L (1-a) J = G A4 a (15) (1 -a) + a* Ax. AxT where ?Ck is a 2-dimensional vector for each of the driver control actions and its derivative (at time instant k), k is the mean (which may he continuously LLpdated during each drive cycle, and reset after each drive cycle), a is a calihraed forgeuing factor, Gk is the estimated inverse covariance matrix, J is the identity matrix, F is the estimated covariance matrix, and Ax[ is the transpose of AxA Irom(l2).

The recursively computed determinant of the covariance matrix, det, is given by dec1 = -a dey(1 + a.AXL. G & ) (16) where ii is the size of the vector XL. It provides a measure of the estimated variability of the driver acceleration, braking andlor steering performance relative to a particifiar driver's mean for these parameters. It also provides a single dimensional measure of total variance which may be tracked 1 0 to capture significant changes in aggregated variability of driver control actions.

The final DCA Index may be scaled to a continuous signal between 0 and 1 and he given by DCA Index = max(Accel Fed Variance, Brake Fed Variance, Steering Variance) (17) Accelerator pedal position as plotted in Figure iSA and steering wheel angle as plotted in Figure 151-3 have been analyzed using (he techniques above. Figure 15C shows example output Rr (he DCA Index based on the input of Figures 1 5A and I SB. ftc determinant of the covariance matrix (16) provides a measure, in this example. of the estimated variability of the 2 0 driver acceleration and steering performance. The respective variabilities have been normalized and aggregated by taking their maximum values to yield the DCA Index as plotted in Figure 1SC.

Vehicle speed is plotted in Figure 15D for reference. Increased variability is captured on the DCA Index scale as values closer to I (indicative of higher driving demand), while decreased variability is captured on the scale as values between, for example, 0 and 0.2 (indicative of low driving demand).

V. Instrument Panel Index Driver interaction with the instrument panel andlor other touch/voice related interfaces may provide an indication of driver activity. An increase in such driver activity level may increase (he cognitive demands on (he driver. As indicated in Fable 1, an increase in driver button pressing activity may increase the driver workload. The frequency of interaction with cabin controls including the wiper control, climate control, volume control, turn indicator, center stack console, window control, power seat control, voice command interface, etc. may be aggregated into a composite index. The Instrument Panel (W) Index thus provides a continuous output (between 0 and 1) representing the interaction of the driver with the instrument panel, electronics, andlor any other HMI.

When a hutton/interface device is pressedJengaged at any time instant k, for example, the output is given by, (18) 1 0 When a button/interface device is not pressed/engaged, the output is given by B13(k)=a-B13(k-1)+( -a) 0 (19) where B?1 is the button/interface pressed/engaged tracking value for each button/inteitace being tracked, and a is a calibratahie forgetting factor.

1'he IP Index output may then he given hy IF Index = max(B?1,BF., ,8P31,BF4 B?,) (20) 2 0 where n is the number of buttons/interfaces being tracked. The IP index may also be determined using any of the aggregation techniques described herein. As an example, techniques similar to those described with reference to (28) and (29) below may be used, etc. I xample turn indicator and air conditioning activity inputs are plotted respectively in Figures 16A and 16B. The resulting lIP Index is dctcrniincd according to (18), (19) and (20) and plotted in Figure 1ÔC. The rise time and steady state value, in this example, are based on the duration of the activity.

VI. headway Index The Headway index provides a continuous variable between 0 and 1 and indicates how close the vehicle heing driven is to the vehicle (or other object) in front of (or beside) it. As indicated in l'able I, increased workload load maybe inferred from reduced mean time headway and/or reduced minimum headway.

Current velocky dependent headway may he ohtained from HW = ( (k) -r (k)) (21) Vf(k) where r(k) is the position of the preceding vehicle at any time instant k, rj(k) is the position of the following vehicle and v/k) is the velocity of the following vehicle. The mean headway, HW,Jk).

may be obtained from HWM (k) = HW,4 (k -1) + a(HW,7 -HW (k -1)) (22) where a is a time constant for exponential filtering, which may he selected as desired. The I-lw Index may then obtained 1mm NW Index = -HW (23) HW) where y is the 11W Index sensitivity gain and H W1vv1x is a calibrated value. The gain may be chosen/adapted depending on the headway time required to meet a maximum index of 1.

The sensitivity gain, in other embodiments, may be chosen/adapted based on, for example, driver type. If a driver type such as "young," "old," "teen," "novice," expert," etc. is known, the sensitivity gain may he adjusted accordingly. A driver may he identified as "young," 2 0 "old," "teen," etc. based on a token carried by them as known in the art, the token may he detected by the vehicle and used to identify the type of driver. Alternatively, the vehicle may provide a select button that lets the driver identify themselves by type. Any suitable/known technique for classifying a driver by type, however, may be used. The sensitivity gain may be increased for "teen" and "novice" drivers, while the sensitivity gain maybe decreased for "expert" drivers. etc. The sensitivity gain, in other embodiments, may be selected to be higher for "teen" and "novice" drivers and selected to be lower for "expert" drivers, etc. hence the 11W Index, given the same headway, may be higher for a "teen" driver and lower for an "expert" driver, etc. Alternatively (or in addition to), the sensitivity gain may be chosen/adapted based on environmental conditions. Wet or icy road conditions, determined by any suitable/known technique such as through the detection of wheel slip, may result in the sensitivity gain being increased. Dry road conditions may result in (he sensitivity gain being decreased. Any suitable environmental conditions including traffic density. geographic location. etc. maybe used to select/alter the sensitivity gain.

headway proximity to infrastructure including intersections, roadwork, high-demand roadway geometry, etc. may also he similarly computed as in (21), (22) and (23). In such cases, the 11W Index output may be given by HW Index = max(HW, HW2,.. . . . .HWJ (24) where n is the number of headway proximity items of high-driving demand being tracked. A weighted function for equation (24) may also be used.

Increased headway returns from increased traffic in adjacent lanes may be used as a bias input to the I-lW Index in other embodiments. (Increased traffic density may increase driving demand as indicated in fable 1.) The time-to-collision, in still other embodiments, may be tracked in the regime of less than 1000 ms. In potentia' imminent crash conditions, (he HW Index output may default to the max value of].

Referring to Figure 17, the time to collision. r. maybe computed as follows --v ±(%ç)2 +2(k)(X) -(A) (25) x or /.=-v,, where V1 is the closing velocity, A1 is the relative acceleration, and Xis the distance between the vehicles. The distance and closing velocity information may be obtained from any suitable/known radar system, vision system. lidar system, vehicle-to-vehicle communication system, etc. Considering the computation of the HW Index in an example vehicle following scenario, Figures 18 through 20 show the host vehicle speed, inter-vehicle closing velocity and range during the scenaro. Figures 21 and 22 show the mean headway (as computed via (22)) and the 11W Index (as computed via (23)), respectively.

VU. Rule-Based Sub-System Referring again to Figure 1, the rule-based sub-system 12 may include a knowledge base and facts for determining an event binary output flag. The sub-system 12 may provide specific expert engineering and vehicle-driver-environment interaction rules to supplement the other components of the system 10. the knowledge maybe represented as a set of rules. Specific activation of vehicle systems may be incorporated.

1 0 Each rule specifies a recommendation of the output workload, and has the IF (condition), TI TEN (action) structure. When the condition part of a rule is satisfied, the action part is executed. Each rule may specify a recommendation of the output workload (0 or I). A number of vehicle parameters including longitudinal acceleration, lateral acceleration, deceleration, steering wheel angle, button usage, etc. (see, e.g., Tables 2a and 2b) may he monitored/obtained in any suitable/known fashion by the sub-system 12 from, for example, the vehicle's CAN bus. Facts associated with these parameters and their combination may be used to set the conditional rules.

A general nile implemented by the sub-system 12 maybe of the form. If

Vehicle -parameter 1)x1 and Vehicle -parameter 2) Then (26) li/f condition is satisfied Event -Flag = [9 otherwise Specific delays or restriction of infotainment or vehicle cabin systems during events are enabled from the expert rules. l'he tide-based output may he further processed to provide a rdative output aggregation based on the usage of a specific feature and the expert notion of the driving demand for the condition.

Ru'es may he based on the information, for example, hsted in tables 2a and 2h above. For example. if steering wheel angte> 105 degrees, then Eveni_Plag = 1. Other rules may also, of course, be constructed.

VHI. Aggregation One or more of the 11W Index, DCA Index, IP Index, and IlL Index may he aggregated by the sub-system 14 to form a Tracking (I) Index using techniques described below.

In embodiments where only one index is usedlcomputedldetermined however, no aggregation may he necessary.

In certain embodiments, short-term aggregation may be used to schedule/delay/suppress information/tasks to be communicated to the dryer. For conditions where the highest driving demand assessed is required. the I' Index maybe given by 1 Index = max (ncA Index, IP Indx. HI. Index, HW Index) (27) In other embodiments, a context dependent aggregation is employed for mean/max output combinations of the index values as described below. With reference to Figure 1 for example, the DCA Index, IP Index, ilL Index, and HW Index may be combined by the sub-system 14 to form a 1 Index given by WI);, Tlndex = (28)

WI

where w1 arc context dependent weights depending on the driving demand value placed on the 2 0 input. Expansion of (28) yields Tindex = WLEDGAWDC +WLE7w + WLE7w7 + WLEw11 +bias (29) + IP + HL + Max( racking -Index) = 1.0 where WLE04, WLE,, WLEH, and WLE arc the DCA Index, IP Index, IlL Index, and 11W Index outputs respectively. The corresponding weights arc given by WDcA, Wjp, WHL and wJ]-w Tables 3 and 4 list example rules for aggregation.

Table 3

Example Rules for Context Based Aggregation Rule If DCA If U' If HI. If HW Then Calculate WLE Index; Index is Index is Index is Index is Max=1.0 Min=0.0; _______ _________ _________ _________ _________ If night, then bias = + 0.2 1 Hi Hi Hi Hi Mean of 4 Indices: w(vector) = [11111 2 Iii Iii Ili Low Mean of 3 of Indices: w(vector) = [111 01 3 Hi Hi Low Low Mean of 2 Indices: w(vector)=[l 1001 4 I-li Low Low Low Max: w(vector) =10 1 001 Low IL Low Low Max: w(vector) =[0 1 0 0] 6 Low Low Hi Low Max: w(vector) =10 0 1 01 7 Low Low Low IL Max: w(vector) =[0 00 1]

Table 4

More Example Rules for Context Based Aggregation Rule If DCA If IP If HLM If HWM Then Calculate WLE Index; Index is Index is Index is Index is Max=i.0 Min=0.0; If night, then bias = + 0.2; If high traffic condition, (hen ______ ________ ________ ________ ________ bias = + 0.2 8 1 nw I ow I nw I w Mean of 4 Indices: w(vector) = [11111 9 Low Hi I-li Hi Meanof3oflndices: w(vector) = [01 1 1] Low Low Ili Ili Mean of 2 Indices: w(vector) = [00 ii] 11 Hi Low Hi Low Mean of 2 Indices: w(vector) = [1 0 1 0] 12 Low Hi Low Hi Mean of 2 Indices: w(vector) = [0 1 0 1] 13 Low Hi Hi Low Mean of 2 Indices: w(vcctor) = [0 1 1 0] 14 Hi Low Hi Hi Mean of 3 of Indices: w(vector) = [1 0 111 Hi Hi Low Hi Mean of 3 of Indices: w(vector) =[1 1 0 1] 16 Iii Low Low Ih Mean of 2 hidices: w(vector) =[1 00 1] The sub-system 16 may use the techniques described above with reference to the sub-system 14 to aggregate the Rule-Based Index and T Index into the WLE Index. As an example, the WLE Index may be give by: W[E Index = max(T Index, Rule -Based Index) (30) An example Rule-Based Index, IP Index, and DCA Index are plotted respectively in Figures 23A through 23C. These indices have been aggregated using the techniques described 1 0 herein and plotted in Figure 23D for conditions where the highest driving demand situations assessed are considered. Vehide specd is plotted in Figure 23N for rcference.

IX. Long Term Characterization The WLE Index, in other embodiments, may be characterized over time to provide for HMI recommendations by (he sub-system 16 and/or dispatcher 18 (depending on the configuration). Long-term WLE characterization may enable IIMI to be tailored to the driver based on the driving demand overtime. Consider, for example, that r is a variable reflecting the WIE Index value for the driver (at any time instant k). Assume the driving demand is categorized into 3 classes as in a,b, c} with fuzzy membership functions,p,jç as defined in Figure 24. Then, 2 0 the driving behavior. dk. can be inferred from the following example computation = Luk)/4k)I (31) If for example i has a value of 0.4, then dk may be represented as [0.18, 0.62, 0] (according to Figure 24). The filtered (long term) version of the driving behavior, df, can be expressed as in the following df = (i -a)df + k (32) where a is a calibratable forgetting factor (which thus specifies/determines (he time period during which the long term version of the driving behavior, d1, is evaluated). ftc long-term probability for each of the classes, (p), may he obtained from (), = ( (33) ic (ub According to (33), the filtered version of the driving behavior for each of the classes. ki1. is divided by the sum of the filtered version of the driving behavior for all of the classes, (df).9. If Ilir example d. is represented as 10, 0.16, 0.381, then (Pja wouM he equal to je{üb.c} 0 divided by 0+0.16+0.38 ((Pk) would be equal to 0). (Pk)h would be equal to 0.16 divided by 0+0.16+0.38 ((Pk)h wouki he equal to 0.29), and (Pk) would he equal to.38 divided by 0+0.16+0.38 ((pt)c woffid he equal to 0.71).

l'he final long-term Wit Index characterization of driving demand, i,may then he inferred from the following = arg max (P& ) (34) ieja.b.c} Using the exampth above, the maximum of the (Pk). values is 0.71 ((Pk)c) Hence, it maybe inferred from (34) that the driving behaviour is currently in the "high demand" class.

X. Dispatcher The dispatcher 18 may apply the computed WLE Index, the long teim characterization of the Wit Index, or anyone of the DCA Tndex, IP Index, HI. Index, and HW Index (in embodiments where only a single index is used/computed/determined) to mod&ate the interaction between the infotainment and/or other dialog systems and the driver. The WLE Index provides the estimated workload load used to set/avoid/tailor/limit/schedule voice commands and other tasks to he presented to the driver to improve functionality and safety.

lxample interaction with the driver may include generating text-to-speech tell-taths, generating avatar communications, generating notifications regarding in-coming phone calls, generating proactive powertrain commands, generating proactive voice recommendations, generating a tactile response via, for example, a tactile steering wheel, or generating other audio, visual and/or tactile outputs, etc. Each of these example driver interface tasks may have a priority associated with it. For example, generating a notification regarding an in-coming phone call may have a high priority whereas generating a proactive voice recommendation may have a low priority.

Any suitable/known technique may be used to assign a priority type to a given driver interface task. As an example, the dispatcher 18 may implement a high/low priority convention wherein all notifications to he generated regarding in-coming phone calls are assigned a high priority and all vehicle initiated recommendations to be communicated to the driver are assigned a 1 0 low priority. Other priority schemes, however, may be used. As an example. numbers between 0 and 1.0 may represent the priority of a task: certain tasks may be assigned a priority of 0.3 while other tasks maybe assigned a priority of 0.8, etc. In other embodiments, the priority type associated with a driver interface task may be assigned by the controller/processor/subsystem (not shown) that generated the task as known in the art.

Certain embodiments may thus permit a modulated presentation of driver interface tasks based on workload and priority, if for example the WLE index (or any one of the indices as the case may he) has a value between 0.4 and 0.6, the dispatcher 18 may only allow high priority driver interface tasks to be executed. The dispatcher 18 may schedule lower priority tasks for later execution conditioned upon the WLE Index attaining a value less than 0.4. if for example the WLE Index has a value between 0.7 and 1.0. the dispatcher 18 may prevent all driver interface tasks from being executed. During these periods of high workload, the dispatcher 18 may schedule high priority tasks for later execution conditioned upon the WLE Index attaining a value less than 0.7 and schedule lower priority tasks for later execution conditioned upon the WLE Index attaining a value less than 0.4.

Similarly, if the long term driving behavior is characterized as "high demand," certain/all tasks regardless of their priority may be suspended/delayed/scheduled until the long term driving behavior is characterized as "medium demand" or "low demand." Alternatively, if the long term driving hehavior has any probability of being in, for example, the "high demand" class.

certain/all tasks may be suspended/delayed/scheduled until the probability of being in "high 3 0 demand" is zero. Other scenarios are, of course, also possible. In embodiments where a priority type is not used to categorize tasks for example, all tasks may be suspended/delayed/scheduled depending on the inferred workload.

In the case of an in-coming phone call received during periods of high workload, the dispatcher 18 may put the call through to a voice mail system. Once the WLE Index has attained an appropriate value, the dispatcher 18 may generate a notification indicating that a call was received.

The algorithms disclosed herein may be deliverable to a processing device, such as any/all of the systems 12, 13, 14, 16, 18, which may include any existing electronic control unit or dedicated electronic control unit, in many forms including, hut not limited to, information permanently stored on non-wr table storage media such as ROM devices and information alterably stored on writeable storage media such as floppy disks, magnetic tapes, CDs, RAM devices, and other magnetic and optical media. The algorithms may also he implemented in a software executable object. Alternatively, the algorthms may be embodied in whole or in part using suitable 1 0 hardware components, such as Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs). state machines, controllers or other hardware components or devices, or a comhination of hardware, software and firmware components.

Whilst specific embodiments of the present invention have been described above, it will be appreciated that departures fl-cm the desciibed embodiments may still fall within the scope of the present invention.

Claims (16)

1. CLAIMS1. A method for managing a plurality of driver interface tasks comprising: categorizing a dnver workload into a plurality of workload classes representing differing levels of driver workload; for a specified time period, determining a relative frequency with which the driver workload falls within each of the classes; scheduling the tasks for execution based on the relative frequencies; and executing the tasks according to the schedule.
2. 2. The method of claim 1 wherein scheduling the tasks for execution based on the rdative frequencies indudes determining a maximum of the rdative frequencies.
3. 3. The method of claim 1 or 2 wherein the tasks are scheduled according to the maximum of the relative frequencies.
4. 4. l'he method of claim 1, 2, or 3 wherein determining a relative frequency with which the driver workload falls within each of the classes includes determining, for each of the classes, a long term driver workload based on the dnver workload from a current time period and the driver workload from a past time period, and a scale factor that determines the specified time period.
5. 5. The method as claimed in any preceding claim wherein the plurality of driver interface tasks includes at least one of generating an audio output, generating a visual output, and generating a tactile output.
6. (3. A vehicle comprising: at least one processor configured to (i) categorize a driver workload into a plurality of workload classes representing differing levels of driver workload, (ii) for a specified time period, 3 0 determine a relative frequency with which the dnver workload falls within each of the classes, (iii) receive a plurality of driver interface tasks to be executed, (iv) schedule the tasks for execution based on the relative frequencies, and (v) execute the tasks according to the schedule.
7. 7. The vehicle of claim 6 wherein the at least one processor is further configured to determine a maximum of the relative frequencies.
8. 8. The vehicle of claim 6 or 7 wherein (he tasks are scheduled according to the maximum of the relative frequencies.
9. 9. The vehicle of claim 6, 7 or 8 wherein the at least one processor is further configured to determine, for each of the purali(y of workload classes, a long term driver workload based on the dryer workload from a current time period and a driver workload from a past time 1 0 period, and a scale factor that determines the specified time period.
10. 10. l'he vehicle of claim 6, 7, 8 or 9 wherein the p'urality of driver interface tasks includes at least one of generating an audio output, generating a visual output, and generating a recommendation for the driver.
11. ii. A vehicle comprising: at least one processor configured to (i) determine a plurality of scaled parameters each representing a driver workload and each based on one of (a) a level of interaction between a driver and one or more driver-vehicle interfaces, (b) a margin between the vehicle's current handling condition and limit handling condition, (c) a variability of at least one driver control action input, or (d) a headway between the vehicle and another object, (ii) receive a plurality of driver interface tasks to be executed, and (iii) selectively delay or prevent at least some of the plurality of driver interface tasks from being executed based on a maximum or aggregation of the plurality ol scaled parameters.
12. 12. The vehicle of claim 11 wherein the at least one processor is further configured to determine the aggregation of the plurality of parameters based on a weighted average of the plurality of parameters.
13. 13. The vehicle of claim 11 or 12 wherein the at least one processor is further configured to determine a parameter representing a workload of the driver based on whether a predefined set of conditions have been satisfied and to selectively delay or prevent at least some of the plurality of driver interface tasks from being executed further based on the determined parameter.
14. 14. The vehicle of claim 11, 12cr 13 wherein the plurality of driver inteitace tasks includes at least one of generating an audio output, generating a visual output, and generating a tactile output.
15. 15. The vehicle as claimed in any of claims 11 to 14 wherein the plurality of driver interface tasks includes generating a recommendation for the driver.
16. 16. The vehicle as claimed in any of claims 11 to 15 wherein each of the 1 0 plurality of driver interface tasks includes a priority type and wherein the at least one processor is configured to selectively delay or prevent at least some of the plurality of driver interface tasks from being executed further based on the priority type.