Construct the Stable-Driving Region with Different Driver’s Habit . 109

the duration between the onset of deviation and the occurrence for steering-wheel.

Subjects have to move the vehicle’s center back to the cruising lane to wait for the next testing deviation produced by the computer when they have been informed in advance. However, the restarting action is not easy to be determined due to the variation of different driving habits, especially the loose drivers which have a larger spread in lateral position so that the distance between the wheel and lane marker can not exactly fixed in the straight-road driving [69]. Therefore, the algorithm to extract the stable-state driving region must be developed before constructing the drowsiness estimation mechanism.

The standard for stable-state range determination is described as below: (1) the lateral position of lane markers within this region should be close to each other; (2) the TLC is larger; (3) The lateral offsets found by the LDW system in section 5.3 must be situated in this region for a long period.

According to the above properties, first of all, we take the lateral offsets with larger TLC about consecutive N frames processed by the LDW system. Second, by the previous statistics, the mean and standard deviation estimated by them with the clustering method are used to model the stable-state region as a normal distribution.

At last, the updating method is developed to adjust the size and location of the range to the changed driving habit for a driver. The flow chart for stable-state region determination is demonstrated in Fig. 69.

To rapidly and precisely find out the optimal parameters of each normal distribution, we choose k-means to initially classify the statistics of N lateral offsets.

Fig. 69: The flow chart for stable-state region determination.

The error function which determines the clustering center point of each group is shown as follows:

In Fig. 69, μ is the mean value of each distribution; δ is the standard deviation of each distribution; w is the weight determined by the probability of each group.

After initializing for each distribution model, we find that the N lateral offsets can be approximately modeled by only three normal distributions, which are respectively located on the points nearby the mean value and 1.5 standard deviations with high probably, as shown in Fig. 70. Therefore, we choose K=3 as the initial clustering numbers.

Not the same as the adaptive background model [70], the human habit can last within a steady behavior style for a long time. Based on this psychological property, we only use a single normal distribution with some update mechanism to model the adaptive stable-state driving region to avoid its unreasonable fluctuation. Updating

0

Fig. 70: The distribution of N lateral offsets and three approximately Gaussian model (N=200 in this Figure.)

the parameters of the stable-state model can adapt to the changed driving habit if the lateral offset is within 2.25 standard deviations of this distribution. The parameters of the distribution which matches the new observation for human habit are updated as follows:

By observing equation (49), the influence for this stable-driving distribution will

be unapparent when the distance between the current lateral offset and the mean value of the model is so far. This property can effectively maintain the stability of this region.

6.2.2 Data Collection and Adjustment for the Realistic Environment

After selecting the suitable driving region for the driver, the experimental statistics evaluated by EEG analysis from BRC can be integrated into our lane departure system. Not the same as experimental condition which stipulated that the reaction behavior can be increasingly slower when the subject starts to enter the drowsy state by observing the trend of reaction time for a long period (about 90 sec), the demand for drowsy estimation mechanism in our system should provide a real-time prediction if the driver is still on the alert. Therefore, we design a gauge chart to estimate and display the current driver’s drowsy degree as much as possible, as shown in Fig. 71 (b).

In Fig. 71 (a), the difference in lateral offset between (B) and (C) is 52.45 pixels, the mean value (A) of stable-driving region is located at pixel value of 123.23, and the reaction time counted from (D) and (E) is 1.65sec, as shown in Fig. 71 (c).

As described in section 6.1.2, the definition of reaction time is the time interval of deviation between the center of the vehicle and that of the cruising lane in the VR-based experimental environment. In other words, the value of deviation can be the same as the lateral offset between the car-body and the lane marker in our vision-based system. By the known stable-driving region determined in section 6.2.1, the drowsiness estimation system can apply to drivers with different driving habits without directly selecting the unchanged center part of ROI, such as the restarting mechanism of BRC. Therefore, the count of reaction time starts when the lateral offset of lane marker deviates outside the stable region, and stops when the driver turns back

the steering wheel exactly in our system. However, since we judge the reactive behavior only by the image contents, the backward motion must be confirmed by the criterion that the direction of the lateral velocity keeps identical until the lateral offset is within the stable region again, as the points (D) and (E) in Fig. 71 (d) separately.

Fig. 71: The mechanism for drowsiness estimation in our LDW system. (a) The relationship between the stable-state region and the lateral deviations. (b) A drowsy-degree gauge chart. (c) A stable-driving

group box, (d) The start and stop points of reaction time.

The flow chart for drowsy degree estimation by the reaction time is shown in Fig. 72.

The discussions about Fig. 72 are described as follows:

(1) The drowsy degree may be subtracted by 10% if the reaction time is never up to 1.5sec for 10sec. This automatic mechanism is based on the VR-based experiment of BRC that the computer will automatically produce the deviation behavior about 5~10sec. After all, the reactive behavior in drowsy state must be

increasingly slower without reducing the alert abruptly.

(2) The variation of drowsy degree displayed in the gauge chart, as demonstrated in Fig. 71 (b), depends on the estimated reaction time of the driver in the realistic environment. To avoid the variances in drowsy degree violating the nature of human operation, the changeful region for each estimation result is limited within plus and minus 20%.

(3) Use the classified alert and drowsy state in Fig. 68 analyzed by EEG-based algorithm as the evidence to determine the cognitive property of the driver in realistic environment.

(4) If the drowsy degree is exceeded 70%, the alarm light with red color will be displayed in our system. On the other words, the alarm light with yellow color will be turned on if the drowsy degree is exceeded 35% but not up to 70%. Otherwise, the green light is showed that the driver is still situated in the safety-state with higher alert.

(5) In general, the lane change maneuver can be not certainly judged as an intentional action for driving or an unintentional behavior with the drowsy consciousness only by the information of deviations. Therefore, the warning mechanism focused on this departing behavior is described as below:

video

Fig. 72 : The flow chart of drowsy degree estimation by the average reaction time evaluated from BRC.

6.3 Summary

As described in section 6.2.1, the straight-road driving distance between the lane marker and wheels can be modeled by the clustered distribution with higher weight and smaller standard deviation. For further adaptation, we develop an update mechanism to make the stable region adaptive to the changeful driving habits of people. Fig. 73 shows the updating process of stable-region described as a statistical chart which contains the information of lateral offsets at the same time. From Fig. 73

(a) to (d), the mean value of the stable-region will increase obviously due to the accumulated lateral offsets which are almost situated over the region and can be regarded as the new driving habit of the driver adequately.

(a) (b)

(c) (d)

Fig. 73: Results of update for the stable-driving region.

(a) (b)

(c) (d)

Fig. 74: Results of the variation of drivers’ drowsy degree by the reaction time.

The relationship between the gauge chart of drowsy degree and the reaction time of drivers is demonstrated in Fig. 74. From Fig. 74 (a) to (b), the reaction time will start to be counted since the lateral offset is outside the stable-region at that moment. Therefore, the drowsy degree can be raised with a specific ratio of the measured reaction time to the threshold which has been evaluated by the EGG-based analysis from BRC. On the other hand, from Fig. 74 (c) to (d), the drowsy degree keeps increasing because the time interval between the current and previous reaction time which are both greater than the threshold is not for 10 sec.