Driver Head Research Articles

Driver Gaze Detection Based on Deep Residual Networks Using the Combined Single Image of Dual Near-Infrared Cameras

Research into the prevention of driver inattention by detecting the driver gaze has become more vital, as traffic accidents due to driver inattention have increased. In a vehicle environment, the conventional gaze-detection methods include detecting the driver gaze using single or multiple cameras. When a single camera is used to detect the driver gaze, excessive rotation of the driver's head may prevent the eye region from being accurately detected, thereby reducing the gaze-detection accuracy. To address this issue, researchers previously attempted gaze detection using dual cameras. However, these methods selectively use the information obtained from each camera; thus, accuracy improvement is limited because the information is not simultaneously used. In addition, the processing complexity increases when images obtained from dual cameras are simultaneously processed. Accordingly, this paper proposes a method to detect the driver's gaze position in the vehicle. This is the first study to calculate the driver gaze via a deep convolutional neural network (CNN) that simultaneously uses image information acquired from the dual near-infrared light cameras. Previous research selectively used one of the images acquired from the dual cameras, and the existing CNN-based gaze-detection methods use multiple deep CNNs for the driver eyes and facial images. However, the proposed method uses one CNN model that integrates all information acquired from the dual cameras into one three-channel image and uses it as an input for the network, thereby increasing the recognition reliability and reducing the computational cost. We conducted experiments based on a self-built driver database that comprised the images from 26 participants (Dongguk dual-camera-based gaze database) and the Columbia gaze dataset, which is an open database. The results demonstrate that the proposed method shows better performance than the existing methods.

Driver head locations: Considerations for head restraint design

Objective: U.S. FMVSS 202a requires that a vehicle head restraint lie within a specified distance (55 mm) from the physical headform on the head restraint measurement device (HRMD). Smaller values of this distance, known as backset, are frequently associated with improved protection against neck injury in rear impact. In some vehicles, small backsets are also associated with complaints of head restraint interference with drivers’ preferred head positions. The objective of this study is to examine head/head restraint distances using data from a lab study of driving posture to provide guidance for safe and comfortable head restraint design.Methods: Head positions were measured for 88 U.S. drivers in a laboratory mockup using a seat from a mid-size sedan. The head restraint was removed to allow measurement of drivers’ preferred head locations without interference from the head restraint. Rates of disaccommodation, defined as interference between predicted possible head restraint locations and drivers’ preferred head locations, were analyzed at HRMD-referenced backsets of 25, 50, 75, and 100 mm measured at 22° and 25° seat back angles.Results: With HRMD-referenced backsets of 25 mm and 50 mm measured at 25°, the head restraint intersected the preferred head locations of 17.9 and 5.2% of the drivers, respectively. An HRMD-referenced backset measured at 22° produced larger accommodation rates than the same backset measured at 25°.Conclusions: The reported distribution of occupant head positions and the resulting restrictions on comfortable head restraint position at various HRMD-referenced backsets and seat back angles help provide guidance for head restraint design. Knowing the actual mean driver-selected seat back angle for a particular vehicle seat and the model presented in this work, a manufacturer can choose a head restraint location that will have a high likelihood of complying with FMVSS backset requirements while also achieving minimal disaccommodation. The findings in this study support the flexibility in the current FMVSS 202a that permits testing at more upright seat back angles than the 25° originally proposed.

드라이버 헤드의 어택각과 타격점 변화가 골프볼의 비거리에 미치는 영향

드라이버 헤드의 어택각과 타격점 변화가 골프볼의 비거리에 미치는 영향

Modular design of fatigue detection in naturalistic driving environments

Research in driver mental fatigue is motivated by the fact that errors made by drivers often have life-threatening consequences. This paper proposes a new modular design approach for the early detection of driver fatigue system taking into account optimisation of system performance using particle swarm optimisation (PSO). The proposed system is designed and implemented using an existing dataset that was simultaneously collected from participants and vehicles in a naturalistic environment. Four types of data are considered as fatigue-related metrics including: vehicle acceleration, vehicle rotation pattern, driver's head position and driver's head rotation. The driver's blink rate data is used in this work as a proxy for ground truth for the classification algorithm. The collected data elements are initially fed to input modules represented by ternary neural network classifiers that estimates alertness. A Bayesian algorithm with PSO is then used to combine and optimise detection performance based on the number of existing input modules as well as their output states. Performance of the developed fatigue-detection system is assessed experimentally with a small data samples of driver trips. The obtained results are found in agreement with the state-of-the-art in terms of accuracy (90.4%), sensitivity (92.6%) and specificity (90.7%). These results are achieved with significant design flexibility and robustness against partial loss of input data source(s). However, due to small sample size of dataset (N = 3), a larger dataset need to be tested with the same system framework to generalise the findings of this work.

드라이버 클럽 백스윙 시 어깨회전, 골반 회전 각도가 헤드스피드에 미치는 영향

This study set out to measure and analyze the effects of shoulder and pelvis rotation angles on head speed in a back swing, which is known to have close connections with the driver head speed and club distance, with ten superior male and female golfers each that were active in League 1 and 2 tours of Korea Professional Golf Association and Korea Ladies Professional Golfers Association. The findings led to the following conclusions: 1. In a driver club back swing, the male and female subjects recorded a mean shoulder rotation angle of 108.73°±9.59° and 115.73°±6.05°, respectively. The female golfers had a greater shoulder rotation by about 7° on average in a back swing than their male counterparts. The male and female golfers recorded a mean pelvis rotation angle of 54.65°±6.71° and 54.75°±7.90°, respectively. There were few differences in the pelvis rotation angles between the male and female golfers. 2. The present study analyzed correlations among the shoulder rotation angle, pelvis rotation angle, and head speed of the male golfers in a back swing and found no significant correlations among them. The female golfers showed significant negative correlations between their shoulder rotation angle and head speed(r= -.786, p<.01).

A study on bus safety operation system applying IoT technology

Background/Objectives: It is a bus safe operation system applying IoT (Internet of Thing) technology. It is the starting point of future economic activitiessuch as past simple transportation methods and various information such as health care, wearable, IoT, We want to change the paradigm into service exchange space. Vehicle infotainment technology utilizes a bi-directional head-up display (HUD) interface technology to detect the wheel state of the steering wheel operation, which is the driver's operation behavior, and the pupil watching forward, and warns if momentary carelessness or continuous carelessness occurs, Avoid.Methods/Statistical analysis: Since the ratio of the eye area in the face area is very small, there is a possibility that the calculation time is greatly increased. In order to optimize the module, based on the analysis of the Korean face information, Thereby reducing the search time. Driver Monitoring System (DMS) developed by Toyota is an infrared camera module installed in front of the dashboard to monitor driver's eyelids and head movements during driving to determine driver's sleepiness.Findings: Based on the know-how of wireless communication-based healthcare device manufacturing technology, we can develop user-oriented products that can differentiate from the existing IoT healthcare products and develop wireless communication-based healthcare for large- We will develop a care terminal to secure marketability.Improvements/Applications: Developed commercial products focusing on the technology stabilization centered on the large transportation profession, obtained stable sensor information and obtained relatively accurate sensing information without inconvenience for long-term operation or work. Active braking assist is a fully automated control system that stops the vehicle when it detects the risk of a frontal collision. Using three radar beams installed in the vehicle, it detects up to 150m in front of the vehicle, helping to avoid the risk of a frontal collision.

Cochlear Damage Caused by the Striking Noise of Titanium Head Golf Driver.

Objectives. To investigate how mouse cochleae are affected by the striking noise of titanium head golf driver.Methods. Thirty-two BALB/c mice (20–22 g) with normal hearing were used. The impact acoustic stimulus generated by the striking of titanium golf driver head was centered around 4.5 kHz with 120.5 dB sound pressure level. The recorded impact noise was provided to mice in two ways with the same exposure time of 288 seconds: 1,440 repetitions and an impact duration of 0.2 seconds for 2 hours (repetitive noise) or serially connected impact noise for 288 seconds (continuous noise). Auditory brainstem responses were measured at baseline, day 7, and day 14 after exposure. The mice were then sacrificed for histology.Results. Both groups showed statistically significant threshold shifts immediately after noise exposure. Mice in the continuous exposure group, except for those exposed to 32 kHz noise, recovered from threshold shifts 1–2 weeks after noise exposure. However, in the repetitive exposure group, threshold shifts remained for 2 weeks after exposure. The repetitive exposure group had greater hair cell damage than did the continuous exposure group. Structural changes in the stria vascularis were observed in the repetitive exposure group.Conclusion. Overexposure to impact noise caused by hitting of titanium head golf driver may be hazardous to the cochlea, and repetitive exposure may induce greater damage than continuous exposure.

Investigating the correspondence between driver head position and glance location.

The relationship between a driver’s glance orientation and corresponding head rotation is highly complex due to its nonlinear dependence on the individual, task, and driving context. This paper presents expanded analytic detail and findings from an effort that explored the ability of head pose to serve as an estimator for driver gaze by connecting head rotation data with manually coded gaze region data using both a statistical analysis approach and a predictive (i.e., machine learning) approach. For the latter, classification accuracy increased as visual angles between two glance locations increased. In other words, the greater the shift in gaze, the higher the accuracy of classification. This is an intuitive but important concept that we make explicit through our analysis. The highest accuracy achieved was 83% using the method of Hidden Markov Models (HMM) for the binary gaze classification problem of (a) glances to the forward roadway versus (b) glances to the center stack. Results suggest that although there are individual differences in head-glance correspondence while driving, classifier models based on head-rotation data may be robust to these differences and therefore can serve as reasonable estimators for glance location. The results suggest that driver head pose can be used as a surrogate for eye gaze in several key conditions including the identification of high-eccentricity glances. Inexpensive driver head pose tracking may be a key element in detection systems developed to mitigate driver distraction and inattention.

Driver injury in near- and far-side impacts: Update on the effect of front passenger belt use

ABSTRACTPurpose: This is a study that updates earlier research on the influence of a front passenger on the risk for severe driver injury in near-side and far-side impacts. It includes the effects of belt use by the driver and passenger, identifies body regions involved in driver injury, and identifies the sources for severe driver head injury.Methods: 1997–2015 NASS-CDS data were used to investigate the risk for Maximum Abbreviated Injury Scale (MAIS) 4 + F driver injury in near-side and far-side impacts by front passenger belt use and as a sole occupant in the driver seat. Side impacts were identified with GAD1 = L or R without rollover (rollover ≤ 0). Front-outboard occupants were included without ejection (ejection = 0). Injury severity was defined by MAIS and fatality (F) by TREATMNT = 1 or INJSEV = 4. Weighted data were determined. The risk for MAIS 4 + F was determined using the number of occupants with known injury status MAIS 0 + F. Standard errors were determined.Results: Overall, belted drivers had greater risks for severe injury in near-side than far-side impacts. As a sole driver, the risk was 0.969 ± 0.212% for near-side and 0.313 ± 0.069% for far-side impacts (P < .005). The driver's risk was 0.933 ± 0.430% with an unbelted passenger and 0.596 ± 0.144% with a belted passenger in near-side impacts. The risk was 2.17 times greater with an unbelted passenger (NS). The driver's risk was 0.782 ± 0.431% with an unbelted passenger and 0.361% ± 0.114% with a belted passenger in far-side impacts. The risk was 1.57 times greater with an unbelted passenger (P < .10). Seat belt use was 66 to 95% effective in preventing MAIS 4 + F injury in the driver. For belted drivers, the head and thorax were the leading body regions for Abbreviated Injury Scale (AIS) 4+ injury. For near-side impacts, the leading sources for AIS 4+ head injury were the left B-pillar, roof, and other vehicle. For far-side impacts, the leading sources were the other occupant, right interior, and roof (8.5%).Conclusions: Seat belt use by a passenger lowered the risk of severe driver injury in side impacts. The reduction was 54% in near-side impacts and 36% in far-side impacts. Belted drivers experienced mostly head and thoracic AIS 4+ injuries. Head injuries in the belted drivers were from contact with the side interior and the other occupant, even with a belted passenger.

Deep Learning-Based Gaze Detection System for Automobile Drivers Using a NIR Camera Sensor.

A paradigm shift is required to prevent the increasing automobile accident deaths that are mostly due to the inattentive behavior of drivers. Knowledge of gaze region can provide valuable information regarding a driver’s point of attention. Accurate and inexpensive gaze classification systems in cars can improve safe driving. However, monitoring real-time driving behaviors and conditions presents some challenges: dizziness due to long drives, extreme lighting variations, glasses reflections, and occlusions. Past studies on gaze detection in cars have been chiefly based on head movements. The margin of error in gaze detection increases when drivers gaze at objects by moving their eyes without moving their heads. To solve this problem, a pupil center corneal reflection (PCCR)-based method has been considered. However, the error of accurately detecting the pupil center and corneal reflection center is increased in a car environment due to various environment light changes, reflections on glasses surface, and motion and optical blurring of captured eye image. In addition, existing PCCR-based methods require initial user calibration, which is difficult to perform in a car environment. To address this issue, we propose a deep learning-based gaze detection method using a near-infrared (NIR) camera sensor considering driver head and eye movement that does not require any initial user calibration. The proposed system is evaluated on our self-constructed database as well as on open Columbia gaze dataset (CAVE-DB). The proposed method demonstrated greater accuracy than the previous gaze classification methods.

A late and failure of airbag deployment case study for drivers of passenger cars in rear-end collisions

The presented study was directed at two types of airbag miss-deployments: late deployment and non-deployment. Late deployment can be a product of override or underride road traffic accidents. Non-deployment can be a product of technical failure or trigger algorithm’s inability to correctly assume the state of the accident to happen. In order to analyse the phenomena through physical tests, a specialized test device was used for a series of 8 non-deployment tests and a series of 4 airbag firing tests, totalling 12 tests. Acceleration based data was recorded and analysed for the movement of the device part simulating the driver head. High speed video recording was used to analyse the mechanics of airbag deployment and correlate with the acceleration based data. It has been determined, in the limitations of the laboratory testing environment, a significant variation of the time frame for the airbag deployments, despite using similar testing conditions and identical tested products. Also, the initial time frame for airbag deployment delay was overshadowed by other factors such as time to impact.

Eliminating Driving Distractions: Human-Computer Interaction with Built-In Applications

This article proposes a novel smart car demonstration platform, with a focus on the intuitive paradigm of human?computer interaction (HCI). The main objective here is to circumvent driver distraction when manipulating built-in car-appropriate applications (apps) for future smart car development. Two major development directions for a smart car's HCI are investigated and discussed in detail: 1) an in-vehicle user interface, which focuses on utilizing the user interface of a transparent windshield display to keep the driver's head up and eyes focused on the roadway, and 2) an in-vehicle interaction design, which emphasizes an intuitive interaction framework to minimize distractions while driving. This article addresses the major difficulties of each direction and discusses related challenges and solutions to achieve safety and comfort during smart car operation.

Driver Movement Patterns Indicate Distraction and Engagement.

Objective This research considers how driver movements in video clips of naturalistic driving are related to observer subjective ratings of distraction and engagement behaviors. Background Naturalistic driving video provides a unique window into driver behavior unmatched by crash data, roadside observations, or driving simulator experiments. However, manually coding many thousands of hours of video is impractical. An objective method is needed to identify driver behaviors suggestive of distracted or disengaged driving for automated computer vision analysis to access this rich source of data. Method Visual analog scales ranging from 0 to 10 were created, and observers rated their perception of driver distraction and engagement behaviors from selected naturalistic driving videos. Driver kinematics time series were extracted from frame-by-frame coding of driver motions, including head rotation, head flexion/extension, and hands on/off the steering wheel. Results The ratings were consistent among participants. A statistical model predicting average ratings from the kinematic features accounted for 54% of distraction rating variance and 50% of engagement rating variance. Conclusion Rated distraction behavior was positively related to the magnitude of head rotation and fraction of time the hands were off the wheel. Rated engagement behavior was positively related to the variation of head rotation and negatively related to the fraction of time the hands were off the wheel. Application If automated computer vision can code simple kinematic features, such as driver head and hand movements, then large-volume naturalistic driving videos could be automatically analyzed to identify instances when drivers were distracted or disengaged.

Real-world adjustments of driver seat and head restraint in Saab 9-3 vehicles

ABSTRACTObjective: Whiplash-associated disorder (WAD), commonly denoted whiplash injury, is a worldwide problem. These injuries occur at relatively low changes of velocity (typically <25 km/h) in impacts from all directions. Rear impacts, however, are the most common in the injury statistics. Females have a 1.5–3 times higher risk of whiplash injury than males. Improved seat design is the prevailing means of increasing the protection of whiplash injury for occupants in rear impacts. Since 1997, more advanced whiplash protection systems have been introduced on the market, the Saab Active Head Restraint (SAHR) being one of the most prominent. The SAHR—which is height adjustable—is mounted to a pressure plate in the seatback by means of a spring-resisted link mechanism. Nevertheless, studies have shown that seats equipped with reactive head restraints (such as the SAHR) have a very high injury-reducing effect for males (∼60–70%) but very low or no reduction effect for females. One influencing factor could be the position of the head restraint relative to the head, because a number of studies have reported that adjustable head restraints often are incorrectly positioned by drivers. The aim was to investigate how female and male Saab drivers adjust the seat in the car they drive the most.Methods: The seated positions of drivers in stationary conditions have been investigated in a total of 76 volunteers (34 females, 42 males) who participated in the study. Inclusion criteria incorporated driving a Saab 9–3 on a regularly basis.Results: The majority of the volunteers (89%) adjusted the head restraint to any of the 3 uppermost positions and as many as 59% in the top position. The average vertical distance between the top of the head and the top of the head restraint (offset) increase linearly with increasing statures, from an average of −26 mm (head below the head restraint) for small females to an average of 82 mm (head above the head restraint) for large males. On average, the offset was 23 mm for females, which is within a satisfactory range and in accordance with recommendations; the corresponding value for males was 72 mm. The backset tended to be shorter among female volunteers (on average 27 mm) compared to the male volunteers (on average 44 mm). Moreover, the backset tended to increase with increasing statures.Conclusions: Incorrect adjustment of the head restraint cannot explain the large differences found between the sexes in the effectiveness of the SAHR system.

Analysis of driver's head tilt using a mathematical model of motion sickness

It is known that car drivers tilt their head toward the center of a curve. In addition, drivers are generally less susceptible to carsickness than are the passengers. This paper uses a mathematical model to investigate the effect of the head-tilt strategy on motion sickness. It is shown that tilting the head in the centripetal direction reduces the estimated motion sickness incidence (MSI), defined as the percentage of subjects who vomited. In addition, the head movements of both drivers and passengers were measured in a real car. It is also shown that the estimated MSI of the drivers is smaller than that of the passengers. Experimental results presented in previous studies demonstrated that the severity of motion sickness was reduced when passengers imitated the head tilt of the driver. These results strongly suggest that the driver's head tilt reduces motion sickness, and this can be understood as a subjective vertical conflict.

Arduino based real time driver drowsiness detection and mobile alert system using bluetooth

The goal of this propose design is to detect drowsiness in drivers to prevent accidents and to improve safety on the highways. A method for detecting driver drowsiness/sleepiness is developed on Arduino microcontroller board based on Atmega328P for real-time monitoring. The readings taken from different sensors used in this project can be seen in an Android based mobile device. It has an MQ-2 alcohol gas sensor to detect alcohol, if consumed by the driver and a relay circuit to stop him from performing the drive if alcohol consumed. The output from the relay activates the GSM module which sends a message to the driver family. MPU-6050 Micro-Electro-Mechanical-System (MEMS) accelerometer to read the head nodding conditions combined with a gyroscope and a temperature sensor to know the temperature inside the vehicle. The system is also able to monitor the driver physiological conditions such as heart pulse rate, eye blink detection using pulse sensor and eye blink sensor respectively. The alarm will be activated when the driver has consumed alcohol or driver head nods frequently or heart pulse rate goes above the limit or when driver eyes get closed. The microcontroller communicates with the android mobile through HC-05 Bluetooth module connected on the board. The android mobile is provided with the facility to view continuous readings from the sensors connected to the system along with the warning status messages through an android software based application installed in the end device.

A reduced feature set for driver head pose estimation

Evaluation of driving performance is of utmost importance in order to reduce road accident rate. Since driving ability includes visual-spatial and operational attention, among others, head pose estimation of the driver is a crucial indicator of driving performance. This paper proposes a new automatic method for coarse and fine head's yaw angle estimation of the driver. We rely on a set of geometric features computed from just three representative facial keypoints, namely the center of the eyes and the nose tip. With these geometric features, our method combines two manifold embedding methods and a linear regression one. In addition, the method has a confidence mechanism to decide if the classification of a sample is not reliable. The approach has been tested using the CMU-PIE dataset and our own driver dataset. Despite the very few facial keypoints required, the results are comparable to the state-of-the-art techniques. The low computational cost of the method and its robustness makes feasible to integrate it in massive consume devices as a real time application.

Redesign of Transjakarta Bus Driver's Cabin

Ergonomic risk at work stations with type Seated Work Control was one of the problems faced by Transjakarta bus driver. Currently “Trisakti” type bus, one type of bus that is used by Transjakarta in corridor 9, serving route Pinang Ranti - Pluit, gained many complaints from drivers. From the results of Nordic Body Map questionnaires given to 30 drivers, it was known that drivers feel pain in the neck, arms, hips, and buttocks. Allegedly this was due to the seat position and the button/panel bus has a considerable distance range (1 meter) to be achieved by drivers. In addition, preliminary results of the questionnaire using Workstation Checklist identified their complaints about uncomfortable cushion, driver's seat backrest, and the exact position of the AC is above the driver head. To reduce the risk level of ergonomics, then did research to design the cabin by using a generic approach to designing products. The risk analysis driver posture before the design was done by using Rapid Upper Limb Assessment (RULA), Rapid Entire Body Assessment (REBA), and Quick Exposure Checklist (QEC), while the calculation of the moment the body is done by using software Mannequin Pro V10.2. Furthermore, the design of generic products was done through the stages: need metric-matrix, house of quality, anthropometric data collection, classification tree concept, concept screening, scoring concept, design and manufacture of products in the form of two-dimensional. While the design after design risk analysis driver posture was done by using RULA, REBA, and calculation of moments body as well as the design visualized using software 3DMax. From the results of analysis before the draft design improvements cabin RULA obtained scores of 6, REBA 9, and the result amounted to 57.38% QEC and moment forces on the back is 247.3 LbF.inch and on the right hip is 72.9 LbF.in. While the results of the proposed improvements cabin design RULA obtained scores of 3, REBA 4, and the moment of force on the back is 90.3 LbF.in and on the right hip is 70.6 LbF.in. This indicated improvement cabin design can reduce ergonomic risk with lower scores on several parts of the body.

Design and Implementation of a Driving Assistance System in a Car-like Robot When Fatigue in the User is Detected

In this paper it is presented a driving assistance system when drowsiness is detected in the driver; the system is tested by a car like robot that is wirelessly controlled by a computational interface developed in Visual Studio 2010, which emulates an automobile panel. Through an artificial vision system the drivers head orientation is monitored for determining if he/she is in a drowsiness state; if so, the robot control turn into automatic and the robot pull over to the right side of the way (built track).

Active control analysis of passenger vehicle interior noise produced from tyre/road interaction

Active noise control (ANC) system is a possible solution for reducing the overall vehicle interior noise and minimising the noise annoyance experienced by the driver and passengers. ANC systems can be installed in vehicles in order to adapt the road noise to customer preferences and expectations. They bring human perception criteria other than loudness into the control design. However, the analysis in this paper deals with applying ANC technology to the tyre/road noise in a vehicle enclosure. In the automotive industry, both cost/weight requirements and increasing comfort demands explain the interest shown at the moment for ANC. In consideration of requirements of quietness, improvement and reduction of weight, an ANC system is based on vibration measurement (reference) at the driver's legs region and sound pressure level at the driver's head position. The neuro-fuzzy technique identifies a specified desired output from two input parameters. The results obtained indicate the feasibility of applying the methodology in this instance. Moreover, it has been verified that a good noise reduction performance can be obtained at the driver's head position over frequencies up to 400 Hz.