Estimated crash avoidance with the hypothetical introduction of automated vehicles: a simulation based on experts’ assessment from French in-depth data

Objective We aimed to quantify, through simulations using real crash data, the number of potentially avoided crashes following different replacement levels of light vehicles by level-5 automated light vehicles (AVs). Methods Since level-5 AVs are not on the road yet, or are too rare, we simulated their introduction into traffic using a national database of all fatal crashes and 5% of injury crashes observed in France in 2011. We fictitiously replaced a certain proportion of light vehicles (LVs) involved in crashes by level-5 AVs, and applied crash avoidance probabilities estimated by a number of experts regarding the capabilities of AVs depending on specific configurations. Estimates of the percentage of avoided crashes per user configuration and according to three selected (10%, 50%, 100%) replacement levels were made, as well as estimates taking into account the relative weight of these crash configurations, and considering fatal and injury crashes separately. Results Our simulation suggests that a reduction of almost half of fatal crashes (56%) and injury crashes (46%) could be expected by replacing all LVs on the road with level-5 AVs. The introduction of AVs would be the least effective for crashes involving a vulnerable road user, especially motorcyclists. Conclusion This result represents encouraging prospects for the introduction of automated vehicles into traffic, while making it clear that, even with all light vehicles replaced with level 5-AVs, all issues would not be solved, especially for crashes involving motorcyclists, cyclists and pedestrians.

A ‘Hands on’ Public Service Program to Help People Stay Sober and Safer on the Roadway

Despite the existence of many different “Don’t drink and drive” programs and campaigns over the past 30 years, alcohol intoxication has continued to account for approximately one quarter to one third of all traffic crashes and crash-related deaths in the United States. The present study describes a new ‘hands on’ evidence-based approach involving real alcohol-intoxicated subjects using a virtual reality (VR) driving ‘game’ to educate the public more effectively about the dangers of drunk driving. A single demonstration subject ‘drove’ a VR-based portable driving simulator on multiple occasions before (Pre) and at 30 min intervals for up to six hours after either vehicle (no alcohol), two, four or six ‘drinks’ (3, 6, or 9 ounces of 80 proof vodka). The defensive driving task was a choice reaction crash avoidance steering maneuver in which the driver’s task was to determine which way to turn to avoid a crash and then aggressively steer away to avoid a crash. The primary dependent variable was the latency to initiate an avoidance steering response. Blood alcohol concentration (BAC) determinations (estimations) were conducted immediately prior to driving tests using BAC Track portable breathalyzers. Control drives (Pre-Treatment and Vehicle treatment) were characterized by an approximately 300–320 ms reaction time to initiate a crash avoidance. Alcohol increased crash-avoidance reaction time. Peak BAC values were 35, 78 and 120 mg/dL for two, four and six drinks, respectively; the decline in BAC was comparable and linear for all three treatments. There was a strong correlation (r = 0.85) between pre-drive BAC level and reaction time across all of the alcohol-related drives. There was a significant increase in crash avoidance reaction time when the BAC was 50–79 mg/dL, which is below the legally defined BAC limit (80 mg/dL) currently used in most states in the US. These results demonstrate that (1) this VR-based driving simulator task could be a useful ‘hands on’ tool for providing public service demonstrations regarding the hazards of drinking and driving and (2) a BAC concentration of 50 mg/dL represents a reasonable evidence-based cut-off for alcohol-impaired driving.

Evidence for the Crash Avoidance Effectiveness of Intelligent and Connected Vehicle Technologies

The Intelligent and Connected Vehicle (ICV) is regarded as a high-tech solution to reducing road traffic crashes in many countries across the world. However, it is not clear how effective these technologies are in avoiding crashes. This study sets out to summarize the evidence for the crash avoidance effectiveness of technologies equipped on ICVs. In this study, three common methods for safety benefit evaluation were identified: Field operation test (FOT), safety impact methodology (SIM), and statistical analysis methodology (SAM). The advantages and disadvantages of the three methods are compared. In addition, evidence for the crash avoidance effectiveness of Advanced Driver Assistance Systems (ADAS) and Vehicle-to-Vehicle communication Systems (V2V) are presented in the paper. More specifically, target crash scenarios and the effectiveness of technologies including FCW/AEB, ACC, LDW/LDP, BSD, IMA, and LTA are different. Overall, based on evidence from the literature, technologies on ICVs could significantly reduce the number of crashes.

Testing the Robustness of Commercial Lane Departure Warning Systems

Assessment methods are needed to rate the performances of advanced driver assistance systems in a range of real-world conditions, enabling the possibility of mandating minimum performance requirements beyond standardized, regulatory pass-or-fail tests, and ultimately ensuring a real and objectively measurable safety benefit. To bridge the gap between regulatory and real-world performance, this work presents a novel robustness assessment methodology and defines a robustness index determined from regulatory tests to analyze the real-world performance of lane departure warning (LDW) systems. In this context, a robust system means that it is insensitive to changes in driving conditions or environmental conditions. Distance to line (DTL) and time to line crossing (TTLC) were calculated for a light truck and a passenger car, and a black box model of the LDW systems was developed to predict their performance over different lane markings, drifting directions, and vehicle lateral and longitudinal speeds. During the test, neither of the vehicles triggered warning in around 10% of the trials despite the perfect condition of the markings painted on the proving ground. The type of lane marking significantly influenced DTL for both vehicles. For the light truck, the drifting direction, marking type, and their interaction were found to be statistically significant, which resulted in a lower robustness index than that of the passenger car. For both vehicles, TTLC was inversely proportional to the lateral speed, which greatly influences crash avoidance.

Collision Avoidance Using Deep Learning-Based Monocular Vision

Autonomous driving technologies, including monocular vision-based approaches, are in the forefront of industrial and research communities, since they are expected to have a significant impact on economy and society. However, they have limitations in terms of crash avoidance because of the rarity of labeled data for collisions in everyday traffic, as well as due to the complexity of driving situations. In this work, we propose a simple method based solely on monocular vision to overcome the data scarcity problem and to promote forward collision avoidance systems. We exploit state-of-the-art deep learning-based optical flow and monocular depth estimation methods, as well as object detection to estimate the speed of the ego-vehicle and to identify the lead vehicle, respectively. The proposed method utilizes car stop situations as collision surrogates to obtain data for time to collision estimation. We evaluate this approach on our own driving videos, collected using a spherical camera and smart glasses. Our results indicate that similar accuracy can be achieved on both video sources: the external road view from the car’s, and the ego-centric view from the driver’s perspective. Additionally, we set forth the possibility of using spherical cameras as opposed to traditional cameras for vision-based automotive sensing.

Road Sign Recognition and Lane Detection using CNN with OpenCV

As a result of road traffic crashes, approximately 1.35 million people die each year, and between 40 to 70 million are injured drastically. Most of these accidents occurs because of to lack of response time to instant traffic events. To develop such recognition and detection system in autonomous cars, it is important to monitor and guide driver through real time traffic events. This involves Road sign recognition and road lane detection. In order to make the driving process safer and efficient, a plan is made to design a driver-assistance system with road sign recognition and lane detection features. In this system we have focused on two important aspects, Road sign recognition and lane detection. The process of road sign recognition in a video can be broken into two main areas for research; detection and classification using convolutional neural networks. Road signs will be detected by analysing colour information, which can be red and blue, contained on the images whereas, in classification phase the signs are classified according to their shapes and characteristics. Along with road sign recognition we also focused on Road Lane detection which is one significant method in the visualization-based driver support structure and capable to be used for vehicle guiding and monitoring, road congestion avoidance, crash avoidance.

A 'Hands on' Public Service Program to Help People Stay Sober and Safer on the Roadway

Background. Despite the existence of many different “Don’t drink and drive” programs and campaigns over the past 30 years, alcohol intoxication has continued to account for approximately one quarter to one third of all traffic crashes and crash-related deaths in the United States. The present study describes a new ‘hands on’ evidence-based approach involving real alcohol-intoxicated subjects using a virtual reality (VR) driving ‘game’ to educate the public more effectively about the dangers of drunk driving. Method. A single demonstration subject ‘drove’ a VR-based portable driving simulator on multiple occasions before (Pre) and at 30 minute intervals for up to six hours after either vehicle (no alcohol), 2, 4 or 6 ‘drinks’ (3, 6, or 9 ounces of 80 proof vodka). The defensive driving task was a choice reaction crash avoidance steering maneuver in which the driver’s task was to determine which way to turn to avoid a crash and then aggressively steer away to avoid a crash. The primary dependent variable was the latency to initiate an avoidance steering response. BAC determinations (estimations) were conducted immediately prior to driving tests using BAC Track portable breathalyzers.Results. Control drives (Pre-Treatment and Vehicle treatment) were characterized by an approximately 300-320 msec reaction time to initiate a crash avoidance. Alcohol increased crash-avoidance reaction time. Peak BAC values were 35, 78 and 120 mg/dl for 2,4 and 6 drinks, respectively; the decline in BAC was comparable and linear for all three treatments. There was a strong correlation (r=0.85) between pre-drive BAC level and reaction time across all of the alcohol-related drives. There was a significant increase in crash avoidance reaction time when the BAC was 50-79 mg/dl, which is below the legally-defined BAC limit (80 mg/dl) currently used in most states in the US.Conclusions. These results demonstrate (1) this VR-based driving simulator task could be a useful ‘hands on’ tool for providing public service demonstrations regarding the hazards of drinking and driving and (2) a BAC concentration of 50 mg/dl represents a reasonable evidence-based cut-off for alcohol-impaired driving.