Left Lane Research Articles

Detecting and Tracking Unsafe Lane Departure Events for Predicting Driver Safety in Challenging Naturalistic Driving Data.

Our goal is to improve driver safety predictions in at-risk medical or aging populations from naturalistic driving video data. To meet this goal, we developed a novel model capable of detecting and tracking unsafe lane departure events (e.g., changes and incursions), which may occur more frequently in at-risk driver populations. The model detects and tracks roadway lane markings in challenging, low-resolution driving videos using a semantic lane detection pre-processor (Mask R-CNN) utilizing the driver's forward lane region, demarking the convex hull that represents the driver's lane. The hull centroid is tracked over time, improving lane tracking over approaches which detect lane markers from single video frames. The lane time series was denoised using a Fix-lag Kalman filter. Preliminary results show promise for robust lane departure event detection. Overall recall for detecting lane departure events was 81.82%. The F1 score was 75% (precision 69.23%) and 70.59% (precision 62.07%) for left and right lane departures, respectively. Future investigations include exploring (1) horizontal offset as a means to detect lead vehicle proximity, even when image perspectives are known to have a chirp effect and (2) Long Short Term Memory (LSTM) models to detect peaks instead of a peak detection algorithm.

Driving Behaviour of Elderly Drivers in Malaysia

Aging of population has increased in most developed countries. In year 2030, the aging population in Malaysia will reach 15% of the population. This situation contributes to the increasing number of older drivers on the road. Thus, it is important to identify the needs and safety aspects of driving among the elderly. The objective of this study was to determine the characteristics of basic driving skills among older drivers and their driving behavior. An instrument adapted from Driver Behavior Questionnaire (DBQ) was used to collect survey data. A total of 240 car drivers aged 50 years and above completed the questionnaire. For observation data, 150 respondents were chosen in order to observe their driving skills especially at intersections and parking areas. A Correlation analysis and Chi-Square test were performed to study the relationship between their age and driving skills. Results showed majority of the respondents demonstrated that they had knowledge of basic driving skills. Observation study at parking areas showed that selection of parking space, in and out parking skills, time taken to park, vehicle position and number of passengers were significant towards gender. While, observation at intersections showed that only vehicle distance was significant towards gender. Three variables (the older drivers being comfortable driving on the left lane, older drivers were often honked during driving and older drivers being slow responding to hazards such as pedestrians at the crossing) were correlated and significant at p<0.05 . This study recommended providing older drivers with priority lanes, special parking areas and car sticker tagging “older driver” in addition to supervising their licenses, conducting driving class sessions and undergoing medical examination for them.

Traffic Flow Modeling of Freeway Variable Speed Limit Control Based on the Big Data of Driving Behavior

Variable speed limit (VSL) control is a flexible restriction on the rate at which motorists can drive on a given stretch of road. Effective VSL control can increase safety and provide clear guidance for motorists. Previous traffic flow models of VSL control were mostly based on the influence of VSL on average speed (macro) or driver’s expected speed (micro). Few models considered the influence of VSL on driver’s actual driving behavior. In this paper, we first briefly introduce the big traffic data involved in this study and explain the mapping relationship between the data and driving behavior. Then, we analyze the driver’s actual driving behavior under the VSL control. Then, an improved single-lane cellular automaton model is established based on the driving behavior characteristics under VSL control. After that, we calibrate the parameters of the single-lane cellular automaton model with the left lane as the calibration object. Finally, this paper uses the proposed single-lane cellular automaton model to simulate the traffic flow characteristics under VSL control. The numerical simulation results show that the simulation of the variable speed limit in different density intervals presents different results, but these results are consistent with the actual situation of variable speed limit control, which verifies the validity of the proposed model.

Takeover Quality: Assessing the Effects of Time Budget and Traffic Density with the Help of a Trajectory-Planning Method

In highly automated driving, the driver can engage in a nondriving task but sometimes has to take over control. We argue that current takeover quality measures, such as the maximum longitudinal acceleration, are insufficient because they ignore the criticality of the scenario. This paper proposes a novel method of quantifying how well the driver executed an automation-to-manual takeover by comparing human behaviour to optimised behaviour as computed using a trajectory planner. A human-in-the-loop study was carried out in a high-fidelity 6-DOF driving simulator with 25 participants. The takeover required a lane change to avoid roadworks on the ego-lane while taking other traffic into consideration. Each participant encountered six different takeover scenarios, with a different time budget (5 s, 7 s, or 20 s) and traffic density level (low or medium). Results showed that drivers exhibited a considerably higher longitudinal and lateral acceleration than the optimised behaviour, especially in the short time budget scenarios. In scenarios of medium traffic density, the trajectory planner showed a moderate deceleration to let a vehicle in the left lane pass; many participants, on the other hand, did not decelerate before making a lane change, resulting in a dangerous emergency brake of the left-lane vehicle. In conclusion, our results illustrate the value of assessing human takeover behaviour relative to optimised behaviour. Using the trajectory planner, we showed that human drivers are unable to behave optimally in urgent scenarios and that, in some conditions, a medium deceleration, as opposed to a maximal or minimal deceleration, is optimal.

Learning From Naturalistic Driving Data for Human-Like Autonomous Highway Driving

Driving in a human-like manner is important for an autonomous vehicle to be a smart and predictable traffic participant. To achieve this goal, parameters of the motion planning module should be carefully tuned, which needs great effort and expert knowledge. In this study, a method of learning cost parameters of a motion planner from naturalistic driving data is proposed. The learning is achieved by encouraging the selected trajectory to approximate the human driving trajectory under the same traffic situation. The employed motion planner follows a widely accepted methodology that first samples candidate trajectories in the trajectory space, then select the one with minimal cost as the planned trajectory. Moreover, in addition to traditional factors such as comfort, efficiency and safety, the cost function is proposed to incorporate incentive of behavior decision like a human driver, so that both lane change decision and motion planning are coupled into one framework. Two types of lane incentive cost — heuristic and learning based — are proposed and implemented. To verify the validity of the proposed method, a data set is developed by using the naturalistic trajectory data of human drivers collected on the motorways in Beijing, containing samples of lane changes to the left and right lanes, and car followings. Experiments are conducted with respect to both lane change decision and motion planning, and promising results are achieved.

Investigating lane change behaviors and difficulties for senior drivers using naturalistic driving data

Introduction: Lane changes can be a complicated maneuver occurring a dynamic environment requiring the integration of many streams of information. Older drivers may struggle with lane changes which may elevate crash risk. Methods: Real-world lane change behaviors were examined using the Second Strategic Highway Research Program Naturalistic Driving Study database. A total of 393 lane changes were observed for two age groups: middle-aged (30–49), and older (70+) drivers. Results: Older drivers were highly likely to fail to execute an over-the-shoulder glance prior to initiation of a lane change (in 98% of left lane changes and 92% for right lane changes). Older drivers also showed higher rates of OTS glance errors at any point during the lane change in 95% of left lane changes and 86% of right-lane changes. Additionally, older drivers frequently failed to activate the turn signal prior to lane change initiation (60% of lane changes for right changes and 59% for left lane changes). Of the older drivers that made side mirror glances, many occurred after the initiation of the maneuver (46% of left lane changes and 58% of right lane changes) suggesting glances were occurring while changing lanes. Conclusions: Results for older drivers showed that many key glances (particularly side mirror checks) and turn signal actuations observed in the current study occurred after the initiation of the lane change, ostensibly when this action may be too late to gather relevant information and avoid a conflict. Practical Applications: Knowledge of glance patterns during lane changes for older drivers can help older drivers maintain travel mobility as they age. Either through training to reinforce OTS and side mirror glances, or supplemental devices like convex mirrors or oversized rearview mirrors, older drivers can reduce high-risk lane change maneuvers and help older drivers to maintain their mobility and independence longer.

An ensemble deep learning approach for driver lane change intention inference

With the rapid development of intelligent vehicles, drivers are increasingly likely to share their control authorities with the intelligent control unit. For building an efficient Advanced Driver Assistance Systems (ADAS) and shared-control systems, the vehicle needs to understand the drivers’ intent and their activities to generate assistant and collaborative control strategies. In this study, a driver intention inference system that focuses on the highway lane change maneuvers is proposed. First, a high-level driver intention mechanism and framework are introduced. Then, a vision-based intention inference system is proposed, which captures the multi-modal signals based on multiple low-cost cameras and the VBOX vehicle data acquisition system. A novel ensemble bi-directional recurrent neural network (RNN) model with Long Short-Term Memory (LSTM) units is proposed to deal with the time-series driving sequence and the temporal behavioral patterns. Naturalistic highway driving data that consists of lane-keeping, left and right lane change maneuvers are collected and used for model construction and evaluation. Furthermore, the driver's pre-maneuver activities are statistically analyzed. It is found that for situation-aware, drivers usually check the mirrors for more than six seconds before they initiate the lane change maneuver, and the time interval between steering the handwheel and crossing the lane is about 2 s on average. Finally, hypothesis testing is conducted to show the significant improvement of the proposed algorithm over existing ones. With five-fold cross-validation, the EBiLSTM model achieves an average accuracy of 96.1% for the intention that is inferred 0.5 s before the maneuver starts.

Impact of Roadside Fixed Objects in Traffic Conditions

Highway Capacity Manual (HCM) assumes that traffic flow rates are equally distributed between lanes, which is not always the case. Lane distribution and speed are influenced by lateral clearance on the roadsides. In Jordan, the absence of well-marked lanes and poor lane discipline results in under-utilizing of the freeway capacity. The objective of this study is to look into the impact of the presence of roadside objects on lane distribution and speed. Test sections were selected on six-lane freeway segments located in sub-urban areas on tangent highway segments. Speed measurements and distribution counts made for each lane on a directional three-lane segment of the freeway. The results showed that lane distribution significantly varies depending on lateral clearance and traffic. As lateral displacement increases, right-lane-use and left-lane-use increases while the middle-lane use remains almost at the same level. Average speed increases as the lateral clearance increases. The results also showed that average speed and lane distribution for 1.5m lateral clearances are very similar to no obstacle conditions. The impact of an obstacle is more significant on the right lane while the use of the left lane fluctuates with a significant increase if traffic flow rates reach high levels.

Situational influencing factors on understanding cooperative actions in automated driving

Cooperative interacting vehicles are a promising approach in the context of automated driving. To ensure understanding and acceptance of such systems, the underlying mechanisms of human cooperation in the context of traffic must be understood. In a driving simulator study, we investigated how situational factors influence cooperative behaviour in a lane change situation on a two-lane German highway during automated driving in the left lane. When another car in the right lane was approaching a slower truck, participants (N = 32) were asked by an automated system to either accelerate, decelerate or maintain speed. The driver’s scope of action, the situation’s criticality for the lane-changing vehicle and the display of intention to change lanes were manipulated. A hierarchical multinomial logistic regression revealed that a wider scope of action, a higher situation’s criticality and signalling the intention to change lanes positively influenced cooperative behaviour by accelerating and decelerating. These results might be applied to design user-centred automated cooperatively interacting vehicles.

Analysis Level of Motorcycle Discipline by Using the Use of Left Flow at Jl. Sudirman Metro City, Lampung

The high number of accidents on Jl. Sudirman Metro city in Lampung is caused by presence of mix traffic, as well as the absence of special treatment for vulnerable motorcycle users, resulting in conflict between motorbikes and other vehicles that have the potential to cause accident. This analysis was carried out to reduce the frequency of high accidents and fatalities due to accident on Jl. Sudirman, an analysis of motorbike lane requirements on Jl. Sudirman and speed management was conducted, one of which was speed limit determination. The analysis samples taken by 7 days survey from 6.00 am – 06.00 pm on Jl Sudirman Metro city Lampung. The result of the surveys showed the criteria for motorcycle lane determination criteria not suitable to implement on Jl. Sudirman segment if it is in accordance with specified indicator because the motorcycle traffic volume is less than 674 smp/hour and 2696 vehicle/hour. This paper indicates that survey analysis on motorcycle driver discipline by using of left flow at Jl. Sudirman Metro city Lampung was not significant. In order to reduce the frequency of high accident we need speed limiting guards with maximum speed 40 km/hours and the signs for motorcycle instruction using the left lane.

Driving maneuver early detection via sequence learning from vehicle signals and video images

Driving Maneuver Early Detection (DMED) is particularly useful for many applications of intelligent vehicle systems, including driver warning and collision avoidance systems. In this paper, we introduce a robust DMED model, denoted as University of Michigan Dearborn (UMD)-DMED, developed using innovative features and deep learning techniques. The UMD-DMED model contains three major computational components, distance based representation of driving context, combined vehicle trajectory features and visual features, and a Long Short-Term Memory (LSTM)-based neural network that captures temporal dependencies of driving maneuvers. To properly evaluate the performances of UMD-DMED, we developed two DMED systems based on the UMD-DMED model, one system is based on partially observed evidence of maneuver events, and another on features observed ahead of the time that driving maneuvers take place. We conducted the extensive experiments using a data set containing 1078 maneuver events extracted from 37 hours of real world driving trips. The results demonstrate that the UMD-DMED model is capable of learning the latent features of five different classes of driving maneuvers, i.e. left turn, right turn, left lane change, right lane change, driving straight. Comparing to four different state-of-the-art DMED systems, the UMD-DMED achieved better detection performances in both, the detection based on partial observations of driver maneuvering, and based on driving context observed ahead-of-time.

Hierarchical reinforcement learning for self‐driving decision‐making without reliance on labelled driving data

Decision making for self-driving cars is usually tackled by manually encoding rules from drivers' behaviors or imitating drivers' manipulation using supervised learning techniques. Both of them rely on mass driving data to cover all possible driving scenarios. This paper presents a hierarchical reinforcement learning method for decision making of self-driving cars, which does not depend on a large amount of labeled driving data. This method comprehensively considers both high-level maneuver selection and low-level motion control in both lateral and longitudinal directions. We firstly decompose the driving tasks into three maneuvers, including driving in lane, right lane change and left lane change, and learn the sub-policy for each maneuver. Then, a master policy is learned to choose the maneuver policy to be executed in the current state. All policies including master policy and maneuver policies are represented by fully-connected neural networks and trained by using asynchronous parallel reinforcement learners (APRL), which builds a mapping from the sensory outputs to driving decisions. Different state spaces and reward functions are designed for each maneuver. We apply this method to a highway driving scenario, which demonstrates that it can realize smooth and safe decision making for self-driving cars.

Connected, Autonomous Cars: Passive Pothole Patrollers

In November 2018, I was driving on the highway at night in upstate New York. The blowing snow and sleet made it difficult to see far ahead, but I noticed several cars pulled over on the side of the road with their hazard lights f lashing. Assuming they had been in an accident, I pulled into the left lane when I suddenly struck an invisible obstacle, blowing my tires. After reaching safe location, I called 911 to relay what little information I could recall. I didn?t know what I had hit or even where it was. That evening, half a dozen motorists struck the same hazard before authorities arrived on the scene.

Design Criteria for Improving Safety Performance of Rural Intersections

This paper, on the one hand, aims to identify significant crash risk factors at unsignalized three-leg intersections connecting rural two-lane two-way roads and minor roads with a STOP control on the approaches (3ST) and, on the other, to make adjustments to the Highway Safety Manual (HSM) procedure, fine-tuning its Safety Performance Function (SPF) based on observation of the local context. Over an 8-year period of study, a total of 240 crashes on 35 3ST intersections were observed, with no geometric-infrastructure adjustments or changes in the Annual Average Daily Traffic (AADT) and surrounding context noted at the intersections investigated. To obtain reliable results, the study period was divided into two groups: (a) 5 years to calibrate a new SPF, and (b) the remaining 3, not included in the first dataset, were used to validate the results. A negative binomial regression model was adopted to calibrate the new SPF. It was found that the AADT on the major and minor roads, the intersection skew angle, the co-occurrence of left and right-turn lanes on the major roads, and lighting seriously affect the crash scenario.

Research on on-street temporary parking effects based on cellular automaton model under the framework of Kerner’s three-phase traffic theory

In this paper, the on-street temporary parking behavior is studied under the framework of Kerner’s three-phase traffic theory based on an improved Kerner–Klenov–Schreckeneberg–Wolf (KKSW) cellular automation (CA) model named KKSW-OSTP (on-street temporary parking) model. The complete parking process is divided into four steps and the microscopic effects of these four steps are analyzed. The parking behavior has significant negative impacts on the traffic flow, while having limited positive impacts on the left lane. The congestion pattern caused by on-street temporary parking is analyzed. Localized synchronized flow pattern (LSP) is observed in low-density traffic condition. Moving synchronized flow pattern (MSP) and General Pattern (GP) are observed in high-density traffic condition. The macroscopic effects of on-street temporary parking on road traffic conditions are analyzed by introducing parking rate into the model. It is found that more on-street temporary parking demand results in higher probability of traffic breakdown and higher congestion level. However, with the increase of parking rate and leave-parking rate, the frequency of rear-end conflicts reduces.

A new lane following method based on deep learning for automated vehicles using surround view images

In this study, a lane following method based on deep learning from surround view images for autonomous driving of a ground vehicle is proposed. Previous methods can be suffered from false detection by hand-craft feature extraction in color-based binarization especially when the surround view images are exposed to unfavorable conditions such as strong shadow, sunlight reflections or shallow puddles on the roads. Thus the proposed method adopts a modified convolutional neural network structure to estimate the 6 coefficients of the left and right lane lines modeled by two quadratic functions from the surround view images of a vehicle. Then, a desired steering wheel angle is calculated using Stanley method to make a test vehicle follow a test lane autonomously by the proposed method. Autonomous driving experiment of the test vehicle using the proposed method was carried out on the test lane with various unfavorable conditions high-curvature lane of a test field. Experiment results showed that the vehicle was self-driven autonomously and stably without any lane departures on the test lane.

Vision Based Driver Alertness System for Lane Detection

This paper presents lane detection used by IP camera. By using the Haugh transform, the lane is detected and gives the notification to the driver. It is also gives the left and right lane marking. In preprocessing, Gaussian filter is apply for smoothing and Canny edge detection method is used as it gives better response than the Sobel, Robert and Marr-Hildreth. Firstly, detect the red edge as left side lane and yellow edge which is for right side lane. The line sharpening is carried out with masking operation. The masking of the image for lane defines the threshold values specifically for Hue, Saturation, and Value (HSV). HSV values are different for red and yellow lane detection. Indication of direction needs to find the vanishing point, and the points are found by using cross product. This paper present display an overview of Hough Transform (HT) calculation as well as utilize it on a path location framework, whereas automobile and mechanical automaton depends upon MATLAB. This paper utilized Internet Protocol (IP) based driving path location framework to accomplish the objective of Advanced Driver Assistance Systems (ADAS). This system is used in car for driverless system.

The crash performance of seagull intersections and intersections with left turn slip lanes

Alternative intersection layouts may reduce traffic delays and/or improve road safety. Two alternatives are reviewed in this research: ‘priority-controlled Seagull intersections’ and ‘priority-controlled intersections with a Left Turn Slip Lane’. Seagull intersections are used to reduce traffic delays. Some do experience high crash rates, however. Left Turn Slip Lanes allow turning traffic to move clear of the through traffic before decelerating, thereby reducing the risk of rear-end crashes. Although there is debate about the safety problems that occur at Seagull intersections and Left Turn Slip Lanes there has been very little research to quantify the safety impact of different layouts. In this study, crash prediction models have been developed to quantify the effect of various Seagull intersection and Left Turn Slip Lane designs on the key crash types that occur at priority intersections. The analysis showed that seagulls are not safe on 4-lane roads, that roadway features like kerb-side parking and nearby intersections can increase crash rates and that left turners in LTSLs can restrict visibility and create safety problems.

Development of Warning System and Driving Model for Overtaking Trucks on Two-Lane Highways

Passing collisions are one of the most serious traffic safety problems on two-lane highways. These collisions occur when a driver overestimates the available sight distance. This paper presents a framework for a passing collision warning system (PCWS) that assists drivers in avoiding passing collisions by reducing the likelihood of human error. The system uses a combination of a camera and radar sensors to identify the impeding vehicle type and to detect the opposing vehicles traveling in the left lane. The study involved the development of a steering control model providing lane-change maneuvers, the design of a driving simulator experiment that allows for the collection of data necessary to estimate passing parameters, and the elaboration of the algorithm for the PCWS based on sensor signals to detect impeding vehicles such as trucks. Simulation tests were carried out to confirm the effectiveness of the proposed PCWS algorithm. The impact of driver behavior on passing maneuvers was also investigated. Mathematical and imitation models were enhanced to implement Simulink for replications of real-life driving scenarios. The different factors that affect system accuracy were also examined.

Active Sensing on a Bicycle for Simultaneous Search and Tracking of Multiple Rear Vehicles

This paper uses an inexpensive single-beam laser sensor mounted on a rotationally controlled platform to simultaneously search for and track multiple vehicles that are behind a bicycle. Vehicles in the bicycle's lane and in the adjacent left lane are both considered. The tasks involved are searching both lanes to detect the presence of vehicles, tracking a vehicle's trajectory once it has been detected, and switching between searching and tracking as needed. A rigorous search algorithm that minimizes the number of sensor rotational angles needed to search the entire region of interest is developed. An error covariance matrix approach is utilized to switch between tracking vehicles and searching the region of interest. Detailed simulation results are presented to show how the developed system handles the absence and presence of vehicles in the two lanes and handles different types of lane change maneuvers while tracking multiple vehicles. Since the developed system uses an inexpensive lightweight sensor that is suitable for on-bicycle implementation, it can be used to detect the danger of a collision and provide a corresponding horn-like audio alert to the motorist.