Search Results

Pre-collision systems (PCS) use forward-looking sensors to detect the location and motion of vehicles ahead and provide a sequence of actions to help the driver either avoid striking the rear-end of another vehicle or mitigate the severity of the crash. The actions include driver alerts, amplification of driver braking as distance decreases (dynamic brake support, DBS), and automatic braking if the driver has not acted or has not acted sufficiently (crash imminent braking, CIB). Recent efforts by various organizations have sought to define PCS objective test procedures and test equipment in support of consumer information programs and potential certification. This paper presents results and insights from conducting DBS and CIB tests on two production vehicles sold in the US. Eleven scenarios are used to assess the systems' performance. The two systems' performance shows that commercial systems can be quite different.

Based on the emergency lane change cases extracted from the China naturalistic driving data, the driving steering behavior divides into three phases: collision avoidance, lateral movement and steering stabilization. Using the steering primitive fitting by Gaussian function, the distribution of the duration time, the relationship between steering wheel rate and deflection were analyzed in three phases. It is shown that the steering behavior essentially is composed of steering primitives during the emergency lane-change. However, the combination of the steering primitives is different according to the specific steering constraints in three phases. In the collision avoidance phase, a single steering primitive with high peak is used for the fast steering; in the lateral movement and stabilization phase, a combination of two or even more steering primitives is built to a more accurate steering.

The effectiveness of Forward Collision Warning (FCW) or similar crash warning/mitigation systems is highly dependent on driver acceptance. If a FCW system delivers the warning too early, it may distract or annoy the driver and cause them to deactivate the system. In order to design a system activation threshold that more closely matches driver expectations, system designers must understand when drivers would normally apply the brake. One of the most widely used metrics to establish FCW threshold is Time to Collision (TTC). One limitation of TTC is that it assumes constant vehicle velocity. Enhanced Time to Collision (ETTC) is potentially a more accurate metric of perceived collision risk due to its consideration of vehicle acceleration. This paper compares and contrasts the distribution of ETTC and TTC at brake onset in normal car-following situations, and presents probability models of TTC and ETTC values at braking across a range of vehicle speeds.

As part of active safety systems for reducing bicyclist fatalities and injuries, Bicyclist Pre-Collision System (BPCS), also known as Bicyclist Autonomous Emergency Braking System, is being studied currently by several vehicles manufactures. This paper describes the development of a surrogate bicyclist which includes a surrogate bicycle and a surrogate bicycle rider to support the development and evaluation of BPCS. The surrogate bicycle is designed to represent the visual and radar characteristics of real bicyclists in the United States. The size of bicycle surrogate mimics the 26 inch adult bicycle, which is the most popular adult bicycle sold in the US. The radar cross section (RCS) of the surrogate bicycle is designed based on RCS measurement of the real adult sized bicycles.

Pedestrian Automatic Emergency Braking (PAEB) for helping avoiding/mitigating pedestrian crashes has been equipped on some passenger vehicles. Since approximately 70% pedestrian crashes occur in dark conditions, one of the important components in the PAEB evaluation is the development of standard testing at night. The test facility should include representative low-illuminance environment to enable the examination of the sensing and control functions of different PAEB systems. The goal of this research is to characterize and model light source distributions and variations in the low-illuminance environment and determine possible ways to reconstruct such an environment for PAEB evaluation. This paper describes a general method to collect light sources and illuminance information by processing large amount of potential collision locations at night from naturalistic driving video data.

According to the U.S. National Highway Traffic Safety Administration, 743 pedal cyclists were killed and 48,000 were injured in motor vehicle crashes in 2013. As a novel active safety equipment to mitigate bicyclist crashes, bicyclist Pre-Collision Systems (PCSs) are being developed by many vehicle manufacturers. Therefore, developing equipment for evaluating bicyclist PCS is essential. This paper describes the development of a bicycle carrier for carrying the surrogate bicyclist in bicyclist PCS testing. An analysis on the United States national crash databases and videos from TASI 110 car naturalistic driving database was conducted to determine a set of most common crash scenarios, the motion speed and profile of bicycles. The bicycle carrier was designed to carry or pull the surrogate bicyclist for bicycle PCS evaluation. The carrier is a platform with a 4 wheel differential driving system.

In China there are many mixed driving roads which cause a lot of safety problems between vehicles and pedalcyclists. Research on driver behavior under risk scenarios with pedalcyclist is relatively few. In this paper driver brake parameters under naturalistic driving are studied and pedalcyclists include bicyclist, tricyclist, electric bicyclist and motorcyclist. Brake reaction time and maximum brake jerk are used to evaluate driver brake reaction speed. Average deceleration is used to evaluate the effect of driver brake operation. Maximum deceleration is used to evaluate driver braking ability. Driver behaviors collected in China are classified and risk scenarios with pedalcyclist are obtained. Driver brake parameters are extracted and statistical characteristics of driver brake parameters are obtained. Influence factors are analyzed with univariate ANOVA and regression analysis.

Roadway departure mitigation systems for helping to avoid and/or mitigate roadway departure collisions have been introduced by several vehicle manufactures in recent years. To support the development and performance evaluation of the roadway departure mitigation systems, a set of commonly seen roadside surrogate objects need to be developed. These objects include grass, curbs, metal guardrail, concrete divider, and traffic barrel/cones. This paper describes how to determine the representative color of these roadside surrogates. 24,762 locations with Google street view images were selected for the color determination of roadside objects. To mitigate the effect of the brightness to the color determination, the images not in good weather, not in bright daylight and under shade were manually eliminated. Then, the RGB values of the roadside objects in the remaining images were extracted.

Objective: Opposite-direction crashes can be extremely severe because opposing vehicles often have high relative speeds. The most common opposite direction crash scenario occurs when a driver departs their lane driving over the centerline and impacts a vehicle traveling in the opposite direction. This cross-centerline crash mode accounts for only 4% of all non-junction non-interchange crashes but 25% of serious injury crashes of the same type. One potential solution to this problem is the Lane Departure Warning (LDW) system which can monitor the position of the vehicle and provide a warning to the driver if they detect the vehicle is moving out of the lane. The objective of this study was to determine the potential benefits of deploying LDW systems fleet-wide for avoidance of cross-centerline crashes. Methods: In order to estimate the potential benefits of LDW for reduction of cross-centerline crashes, a comprehensive crash simulation model was developed.

While the number of traffic fatalities as a whole continues to decline steadily over time, the number of pedestrian fatalities continues to rise (up 8% since 2009) and comprises a larger fraction of these fatalities. In 2011 there were 4,432 pedestrians killed and an estimated 69,000 pedestrian injuries [1]. A new generation of Pedestrian Pre-Collision Systems (PCS) is being introduced by car manufactures to mitigate pedestrian injuries and fatalities. In order to evaluate the performance of pedestrian PCS, The Transportation Active Safety Institute (TASI) at Indiana University-Purdue University Indianapolis is developing a set of test scenarios and procedures for evaluating the performance of pedestrian PCS with the support of the Collaborative Safety Research Center of Toyota. Pedestrian crashes are complex in that there are many aspects about location, driver behavior, and pedestrian behaviors that may have implications for the performance of the PCS.

In order to improve the energy efficiency of hybrid electric vehicles and to improve the effectiveness of energy management algorithms, it is very important to predict the future changes of traffic parameters based on traffic big data, so as to predict the future vehicle speed change and to reduce the friction brake. Under the framework of deep learning, this paper establishes a Long Short-Term Memory (LSTM) artificial neural network traffic flow parameter prediction model based on time series through keras library to predict the future state of traffic flow. The comparison experiment between Long Short-Term Memory (LSTM) artificial neural network model and Gate Recurrent Unit (GRU) model using US-101 data set shows that LSTM has higher accuracy in predicting traffic flow velocity.

Based on the intelligent transportation system, this paper focuses on the control method of the autonomous intersection. First of all, a vehicle scheduling method based on global planning is presented for the active intersection under the environment of sparse traffic. This method is a collaborative control strategy that optimizes the order of vehicles through the intersection. By modeling the vehicle at the intersection, vehicle information and road information are used to set up a control strategy for all vehicles in a specific range, so that all vehicles can be controlled to optimize the global travel time. Finally, we build an intersection simulation experiment platform which is used to simulate the intersection vehicle control strategy. The simulation results show that the proposed method has a good effect on the intersection vehicle control under the sparse traffic environment.

There are over 4,500 fatal intersection crashes each year in the United States. Intersection Advanced Driver Assistance Systems (I-ADAS) are emerging active safety systems designed to detect an imminent intersection crash and either provide a warning or perform an automated evasive maneuver. The performance of an I-ADAS will depend on the ability of the onboard sensors to detect an imminent collision early enough for an I-ADAS to respond in a timely manner. One promising method for determining the earliest detection opportunity is through the reconstruction of real-world intersection crashes. After determining the earliest detection opportunity, the required sensor range, orientation, and field of view can then be determined through the simulation of these crashes as if the vehicles had been equipped with an I-ADAS.

Road departures are one of the most deadly crash modes, accounting for nearly one third of all crash fatalities in the US. Lane departure warning (LDW) systems can warn the driver of the departure and lane departure prevention (LDP) systems can steer the vehicle back into the lane. One purpose of these systems is to reduce the quantity of road departure crashes. This paper presents a method to predict the maximum effectiveness of these systems. Thirty-nine (39) real world crashes from the National Automotive Sampling System (NASS) Crashworthiness Data System (CDS) database were reconstructed using pre-crash velocities downloaded for each case from the vehicle event data recorder (EDR). The pre-crash velocities were mapped onto the vehicle crash trajectory. The simulations assumed a warning was delivered when the lead tire crossed the lane line. Each case was simulated twice with driver reaction times of 0.38 s and 1.36 s after which time the driver began steering back toward the road.

Motor vehicle crashes involving child Vulnerable Road Users (VRUs) remain a critical public health concern in the United States. While previous studies successfully utilized the crash scenario typology to examine traffic crashes, these studies focus on all types of motor vehicle crashes thus the method might not apply to VRU crashes. Therefore, to better understand the context and causes of child VRU crashes on the U.S. road, this paper proposes a multi-step framework to define crash scenario typology based on the Fatality Analysis Reporting System (FARS) and the Crash Report Sampling System (CRSS). A comprehensive examination of the data elements in FARS and CRSS was first conducted to determine elements that could facilitate crash scenario identification from a systematic perspective. A follow-up context description depicts the typical behavioral, environmental, and vehicular conditions associated with an identified crash scenario.

Analyzing dynamic postures of vehicle occupants in various situations is valuable for improving occupant accommodation and safety. Accurate tracking of an occupant’s head is of particular importance because the head has a large range of motion, controls gaze, and may require special protection in dynamic events including crashes. Previous vehicle occupant posture studies have primarily used marker-based optical motion capture systems or multiple video cameras for tracking facial features or markers on the head. However, the former approach has limitations for collecting on-road data, and the latter is limited by requiring intensive manual postprocessing to obtain suitable accuracy. This paper presents an automated on-road head tracking method using a single Microsoft Kinect V2 sensor, which uses a time-of-flight measurement principle to obtain a 3D point cloud representing objects in the scene at approximately 30 Hz.

To minimize the risk of run-off-road collision, new technology in Advanced Driver Assistive System (ADAS), called Road Departure Mitigation Systems (RDMS), is being introduced recently. Most of the RDMS rely on clear lane markings to detect road departure events using the camera for decision-making and control actions. However, many roadsides do not have lane markings or clear lane markings, especially in some rural and residential areas. The absence of lane markings forces RDMS to observe roadside objects and road edge and use them as a reference to determine whether a roadway departure incident is happening or not. To support and guide for developing and evaluating RDMS, a testing environment with representative road edges needs to be established. Since the grass road edge is the most common in the US, the grass road edge should be included in a testing environment.

Road Departure Mitigation System (RDMS) is a new feature in vehicle active safety systems. It may not rely only on the lane marking for road edge detection, but other roadside objects This paper discusses the radar aspect of the RDMS testing on roads with grass road edges. Since the grass color may be different at different test sites and in different seasons, testing of RDMS with real grass road edge has the repeatability issue over time and locations. A solution is to develop surrogate grass that has the same characteristics of the representative real grass. Radar can be used in RDMS to identify road edges. The surrogate grass should be similar to representative real grass in color, LIDAR characteristics, and Radar characteristics. This paper provides the 24 GHz and 77 GHz radar characteristic specifications of surrogate grass.

Based on RADAR and LiDAR measurements of deer with RADAR and LiDAR in the Spring and Fall of 2014 [1], we report the best fit statistical models. The statistical models are each based on time-constrained measurement windows, termed test-points. Details of the collection method were presented at the SAE World Congress in 2015. Evaluation of the fitness of various statistical models to the measured data show that the LiDAR intensity of reflections from deer are best estimated by the extreme value distribution, while the RCS is best estimated by the log-normal distribution. The value of the normalized intensity of the LiDAR ranges from 0.3 to 1.0, with an expected value near 0.7. The radar cross-section (RCS) varies from -40 to +10 dBsm, with an expected value near -14 dBsm.

Animal-vehicle collision (AVC) is a significant safety issue on American roads. Each year approximately 1.5 million AVCs occur in the U.S., the majority of them involving deer. The increasing use of cameras and radar on vehicles provides opportunities for prevention or mitigation of AVCs, particularly those involving deer or other large animals. Developers of such AVC avoidance/mitigation systems require information on the behavior of encountered animals, setting characteristics, and driver response in order to design effective countermeasures. As part of a larger study, naturalistic driving data were collected in high AVC incidence areas using 48 participant-owned vehicles equipped with data acquisition systems (DAS). Continuous driving data including forward video, location information, and vehicle kinematics were recorded. The respective 11TB dataset contains 35k trips covering 360K driving miles.