Limits...
Optical flow and driver's kinematics analysis for state of alert sensing.

Jiménez-Pinto J, Torres-Torriti M - Sensors (Basel) (2013)

Bottom Line: Our experiments show the effectiveness of the approach with a correct eyes detection rate of 99.41%, on average.The results obtained with the proposed approach in an experiment involving fifteen persons under different levels of sleep deprivation also confirm the discriminability of the fatigue levels.In addition to the measurement of fatigue and drowsiness, the pose tracking capability of the proposed approach has potential applications in distraction assessment and alerting of machine operators.

View Article: PubMed Central - PubMed

Affiliation: Department of Electrical Engineering, Pontificia Universidad Catòlica de Chile, Vicuña Mackenna 4860, Casilla 306-22, Santiago, Chile. jejimenp@puc.cl

ABSTRACT
Road accident statistics from different countries show that a significant number of accidents occur due to driver's fatigue and lack of awareness to traffic conditions. In particular, about 60% of the accidents in which long haul truck and bus drivers are involved are attributed to drowsiness and fatigue. It is thus fundamental to improve non-invasive systems for sensing a driver's state of alert. One of the main challenges to correctly resolve the state of alert is measuring the percentage of eyelid closure over time (PERCLOS), despite the driver's head and body movements. In this paper, we propose a technique that involves optical flow and driver's kinematics analysis to improve the robustness of the driver's alert state measurement under pose changes using a single camera with near-infrared illumination. The proposed approach infers and keeps track of the driver's pose in 3D space in order to ensure that eyes can be located correctly, even after periods of partial occlusion, for example, when the driver stares away from the camera. Our experiments show the effectiveness of the approach with a correct eyes detection rate of 99.41%, on average. The results obtained with the proposed approach in an experiment involving fifteen persons under different levels of sleep deprivation also confirm the discriminability of the fatigue levels. In addition to the measurement of fatigue and drowsiness, the pose tracking capability of the proposed approach has potential applications in distraction assessment and alerting of machine operators.

Show MeSH

Related in: MedlinePlus

Pose rate of change and pose computed from the integral of the pose rate without reset conditions for the awake (a); semi-drowsy (b) and drowsy (c) drivers, corresponding to test subjects 1, 6 and 11, respectively.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3673081&req=5

f12-sensors-13-04225: Pose rate of change and pose computed from the integral of the pose rate without reset conditions for the awake (a); semi-drowsy (b) and drowsy (c) drivers, corresponding to test subjects 1, 6 and 11, respectively.

Mentions: Another indicator of drowsiness is the change in the driver's pose. Table 2 presents the average root mean square (RMS) value of the pose magnitude and the pose rate magnitude for the different group of drivers. The pose magnitude is computed as the Euclidean norm of the vector of pose angles (α1, β1, γ1, α2, β2). Similarly, the pose rate magnitude is calculated as the Euclidean norm of the vector containing the time derivatives of the pose angles. The pose magnitude RMS value for the awake and drowsy drivers was similar, and on average, larger than that of semi-drowsy drivers. However, there is a positive correlation between the mean PERCLOS measure and the RMS value of the rate of change in pose. This is consistent with the knowledge that a drowsy driver will attempt to make fast sudden corrections to deviations from the lane, and it is expectable that the driver will also try to regain the sitting pose quickly, while avoiding to fall asleep. It has been argued, see for example [6,12,13], that monitoring corrections in driving maneuvers and pose changes may not provide information sufficiently in advance to warn the driver. In fact, the evolution of the pose of the driver in time does not seem to provide an indication of fatigue as clearly as the rate of change of the pose, according to our results in Figure 12 for a selection of one awake, one semi-drowsy and one drowsy driver. Due to space limitations, it is not possible to include the plots for the fifteen subjects; however, the curves have similar evolutions for drivers within the same group (awake, semi-drowsy or drowsy). The first column of Figure 12 presents the rate of change of the pose angles, α1, β1, γ1, α2 and β2versus time for the awake subject 1 (first row), for the semi-drowsy driver 6 (second row) and the drowsy driver (last row). Clearly, the awake driver presents less sudden rapid motions than the semi-drowsy or the drowsy driver. Integrating the pose rate angles yields the curves in the right column of Figure 12. This second column corresponds to the pose without considering the reset conditions, and therefore, accumulates the measurement errors in the 45 minutes (2,700 seconds) of the experiment. The integrated pose curves for the drowsy driver deviate more from the starting pose than those of the semi-drowsy or fully awake drivers, due to the larger number of sudden pose corrections. We observed that awake drivers tended to seek a more comfortable sitting position after a while of driving or simply changed position because of boredom. However, awake drivers kept their position for longer periods. On the other hand, drowsy driver's were struggling not to fall asleep, seemed also more concerned about not failing the test and, therefore, would move quickly to regain control of themselves.


Optical flow and driver's kinematics analysis for state of alert sensing.

Jiménez-Pinto J, Torres-Torriti M - Sensors (Basel) (2013)

Pose rate of change and pose computed from the integral of the pose rate without reset conditions for the awake (a); semi-drowsy (b) and drowsy (c) drivers, corresponding to test subjects 1, 6 and 11, respectively.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3673081&req=5

f12-sensors-13-04225: Pose rate of change and pose computed from the integral of the pose rate without reset conditions for the awake (a); semi-drowsy (b) and drowsy (c) drivers, corresponding to test subjects 1, 6 and 11, respectively.
Mentions: Another indicator of drowsiness is the change in the driver's pose. Table 2 presents the average root mean square (RMS) value of the pose magnitude and the pose rate magnitude for the different group of drivers. The pose magnitude is computed as the Euclidean norm of the vector of pose angles (α1, β1, γ1, α2, β2). Similarly, the pose rate magnitude is calculated as the Euclidean norm of the vector containing the time derivatives of the pose angles. The pose magnitude RMS value for the awake and drowsy drivers was similar, and on average, larger than that of semi-drowsy drivers. However, there is a positive correlation between the mean PERCLOS measure and the RMS value of the rate of change in pose. This is consistent with the knowledge that a drowsy driver will attempt to make fast sudden corrections to deviations from the lane, and it is expectable that the driver will also try to regain the sitting pose quickly, while avoiding to fall asleep. It has been argued, see for example [6,12,13], that monitoring corrections in driving maneuvers and pose changes may not provide information sufficiently in advance to warn the driver. In fact, the evolution of the pose of the driver in time does not seem to provide an indication of fatigue as clearly as the rate of change of the pose, according to our results in Figure 12 for a selection of one awake, one semi-drowsy and one drowsy driver. Due to space limitations, it is not possible to include the plots for the fifteen subjects; however, the curves have similar evolutions for drivers within the same group (awake, semi-drowsy or drowsy). The first column of Figure 12 presents the rate of change of the pose angles, α1, β1, γ1, α2 and β2versus time for the awake subject 1 (first row), for the semi-drowsy driver 6 (second row) and the drowsy driver (last row). Clearly, the awake driver presents less sudden rapid motions than the semi-drowsy or the drowsy driver. Integrating the pose rate angles yields the curves in the right column of Figure 12. This second column corresponds to the pose without considering the reset conditions, and therefore, accumulates the measurement errors in the 45 minutes (2,700 seconds) of the experiment. The integrated pose curves for the drowsy driver deviate more from the starting pose than those of the semi-drowsy or fully awake drivers, due to the larger number of sudden pose corrections. We observed that awake drivers tended to seek a more comfortable sitting position after a while of driving or simply changed position because of boredom. However, awake drivers kept their position for longer periods. On the other hand, drowsy driver's were struggling not to fall asleep, seemed also more concerned about not failing the test and, therefore, would move quickly to regain control of themselves.

Bottom Line: Our experiments show the effectiveness of the approach with a correct eyes detection rate of 99.41%, on average.The results obtained with the proposed approach in an experiment involving fifteen persons under different levels of sleep deprivation also confirm the discriminability of the fatigue levels.In addition to the measurement of fatigue and drowsiness, the pose tracking capability of the proposed approach has potential applications in distraction assessment and alerting of machine operators.

View Article: PubMed Central - PubMed

Affiliation: Department of Electrical Engineering, Pontificia Universidad Catòlica de Chile, Vicuña Mackenna 4860, Casilla 306-22, Santiago, Chile. jejimenp@puc.cl

ABSTRACT
Road accident statistics from different countries show that a significant number of accidents occur due to driver's fatigue and lack of awareness to traffic conditions. In particular, about 60% of the accidents in which long haul truck and bus drivers are involved are attributed to drowsiness and fatigue. It is thus fundamental to improve non-invasive systems for sensing a driver's state of alert. One of the main challenges to correctly resolve the state of alert is measuring the percentage of eyelid closure over time (PERCLOS), despite the driver's head and body movements. In this paper, we propose a technique that involves optical flow and driver's kinematics analysis to improve the robustness of the driver's alert state measurement under pose changes using a single camera with near-infrared illumination. The proposed approach infers and keeps track of the driver's pose in 3D space in order to ensure that eyes can be located correctly, even after periods of partial occlusion, for example, when the driver stares away from the camera. Our experiments show the effectiveness of the approach with a correct eyes detection rate of 99.41%, on average. The results obtained with the proposed approach in an experiment involving fifteen persons under different levels of sleep deprivation also confirm the discriminability of the fatigue levels. In addition to the measurement of fatigue and drowsiness, the pose tracking capability of the proposed approach has potential applications in distraction assessment and alerting of machine operators.

Show MeSH
Related in: MedlinePlus