Intersection collisions constitute approximately 26% of all accidents in the United States. Unlike longitudinal and lateral collisions that occur in a single direction of traffic flow, most cases of intersection collisions involve vehicles in different directions crossing paths. Because of their complexity and demand on the perceptual and decision making abilities of the driver, intersections present an increased risk of collisions between automobiles.
Intersection Collision Avoidance Systems (CAS) address hazardous situations occurring in the vicinity of intersections. This user service warns a driver of imminent collisions when approaching or crossing intersections that have some form of traffic control. Intersection CAS would provide the driver with assistance in avoiding collisions at intersections due to inattention, faulty perception, obstructed views or intoxication. Specifically, the system will be able to determine if the vehicle is about to violate the right-of-way of another vehicle or disregard the intersection traffic control. If it is determined that the right-of-way is about to be violated, the system will warn the driver of the imminent danger.
The National Highway Traffic and Safety Administration (NHTSA) has sponsored research in the area of Intersection Collision Avoidance and has defined four distinct scenarios that categorize the different types of intersection accidents. These scenarios were derived using accident data from a history of crashes involving intersection collisions. A review of these crashes allowed researchers to construct a taxonomy of accident types. Groupings of similar intersection crash factors yielded the four crash scenarios recognized in the system architecture. These scenarios are shown in Figure 1 and are defined as:
- Parallel Path - Left Turn Across Path
- Perpendicular Path - Inadequate Gap
- Perpendicular Path - Violation of Traffic Control Device (TCD)
- Premature Intersection Entry
Figure 1: Intersection
Collision Scenarios
Intersection Collision Avoidance technology focuses upon minimizing the risk involved when a vehicle nears and/or enters an intersection. By keeping clear of the scenarios mentioned above, a driver can avoid the risk of a serious accident.
Much like the systems that are prevalent in Longitudinal Collision Avoidance, all current work with Intersection CAS is based on radar technology. Yet, intersection collisions generally involve vehicles that are moving at 90 degrees from each other and often at relatively high speeds. The challenge lies in sensing and processing these dangerous scenarios in an acceptable amount of time. The combination of two-dimensional motion, high speeds, large separation distances, and multiple vehicles with the potential for conflict make these systems more complex. To manage these complexities, advanced algorithms are used to examine each scenario and to provide the driver with the most useful information.
The technological solutions being applied to the field of Intersection Collision Avoidance fall into 2 main categories: Vehicle-Based Video Image Processing and Cooperative Infrastructure-Vehicle Communications. Both systems are radar based, but offer different benefits in terms of costs and efficiencies.
Vehicle-based Image ProcessingVehicle-Based Image Processing is a simpler radar-based technology being used to detect, identify, and transmit sign and intersection information to the driver. One type of Vehicle-Based Image Processing is Traffic Sign Recognition. This system is based on the definable patterns of standard traffic signs, which are read in by a radar seeking certain intersection characteristics. Its four basic functions are road-sign detection, content recognition, image acquisition, and characteristic extraction. Although quite simple, Traffic Sign Recognition has the advantage of being extremely quick. It is most useful, however, when performed in conjunction with other Video-Based Image Processing technologies, such as the Intersection Crossing Aid.
The Intersection Crossing Aid is designed to assist drivers across an intersection after stopping at a Stop sign by detecting approaching vehicle. Optical flow is used for detecting the movement of objects in the radar image. An on-board processor then performs multiple types of object recognition in order to warn the driver of any obstructions or possible conflicts. This system is more robust than the Traffic Sign Recognition system, but requires an additional camera and processing unit. Both of these systems are vehicle-based and do not require changes to the existing infrastructure, yet are limited in their performance and in the type of information disseminated to the driver.
Cooperative Vehicle-Infrastructures CommunicationsCooperative Infrastructure-Vehicle Communications require changes to the current infrastructure, involving higher costs, but have demonstrated improved results. These types of systems consist of vehicles continually relaying information to a beacon located in the approaching intersection. This beacon will supply vehicles in all directions with information describing traffic volume, possible conflicts, if there is a signal, etc. Two types of systems are currently being used: Vehicle-to-Roadside IR Communications and Roadside Transponders.
A Vehicle-to-Roadside IR Communications system is focused mainly on blind intersections, and makes use of the beacon technology mentioned above. An on-board transceiver sends vehicle ID information (speed, whether making a left or right turn, etc.) ahead to the beacon, which transmits accumulated data to vehicles in all directions. From this information, on-board computers run complex algorithms to see if the intersection will be safe to cross, or whether or not slowing down or stopping is necessary.
A Roadside Transponder system utilizes Vehicle-to-Roadside communications and allows a transceiver mounted on the vehicle to transmit its own personal characteristics to roadside units spaced equidistant along an interstate or highway. The transceiver can then read in information regarding the appropriate speed that should be taken. Used in conjunction with an Intelligent Cruise Control (ICC) System as discussed in Longitudinal Collision Avoidance section, a Roadside Transponder system can be very accurate, having a standard deviation error of 1 meter in a measurement range of 40 meters.
The Cooperative Infrastructure-Vehicle Communication Systems are much more advanced than the Video Image Processing Systems, yet are costlier and harder to implement on a system-wide level. Both however, show marked advances in the area of Intersection Collision Avoidance and offer insight into the characteristics that a successful Intersection CAS should have. These technologies serve as the basis for research projects and testing around the country.
There are currently many projects underway to develop Intersection Collision Avoidance Systems. Three projects that typify the direction of Intersection CAS research include work done by Calspan SRL Corporation, the Seattle Research Center, and Veridian.
CALSPAN SRL CORPORATION
Researchers at Calspan SRL Corporation are conducting an experiment utilizing both a threat-detection component and an Intersection-to-Vehicle Communications component. The threat detection component consists of two opposed millimeter wave radar systems that rotate 90 degrees to either side of a vehicle, scanning the intersection environment (See Figure 2).
Figure 2: Radar Scanning
of Intersection 
Multiple signal processors are employed to analyze the returned radar data, and a tracking algorithm determines if detected vehicles are on a collision course with the equipped vehicle. The complex algorithms used were derived from real data collected by researchers that detailed the circumstances in which crashes occurred, as well as information related to the drivers intended maneuvers and other critical pre-crash data.
The Intersection-to-Vehicle component uses the beacon technology previously described. The communication equipment will be installed at intersections with controlled traffic signals, and will be able to pass signal phasing information to the approaching vehicle using narrow-band radio frequencies. The driver is then prompted with the range of speeds that will allow him to reach the intersection during the green time. However, this speed will be superseded by any hazardous situations discovered by the Threat Detection component.
The two components of this system cooperate with one another to warn the driver of any possible conflicts in the roadway ahead. The system does not correct for the situation at hand, yet will give an audible warning when possible danger occurs. Presently this concept only allows for one-way communication, from intersection to vehicle. Expansion of the capabilities could provide two-way communication that would allow equipped vehicles to be used for more advanced traffic management functions.
SEATTLE RESEARCH CENTER
The Seattle Research Center has been working on a brake pulsing mechanism
as a haptic warning to possible intersection collisions. The system will
scan the approaching intersection utilizing the millimeter radar technology
previously mentioned, and will attempt to alert the driver through haptic,
or tactile, vibrations on the brake pad. Much of the research involved
in this project has dealt with studying driver inputs regarding the actions
taken prior to a crash. In this way, the optimal time for a haptic alert
can be estimated and applied.
The primary haptic breaking design was tested using criteria established by standard human reaction times. The design uses an actuator, hydraulic calipers, and a pump-valve system that works on the vehicle brake pedal. On-road testing has determined optimal pulsing rate as well as a resulting decrease in velocity of the test vehicle when action is not taken. The haptic break system is fully disengaged when the driver actuates the vehicle brake pad. In test trials, this system used with a standard auditory warning has proven to be quite effective in mitigating intersection collisions.
VERIDIAN OPERATIONS
Veridian Operations is developing an on-vehicle threat detection
system for intersection collision avoidance using a combination of GPS/GIS
technology and radar instrumentation. Through the Global Positioning
System (GPS) and the Geographical Information System (GIS), on-board
computers can identify the roadway on which a vehicle is travelling and
the approach of an intersection. Using a GIS map datafile, the configuration
of the intersection and the angles of the adjoining roadways can be obtained.
This feature allows the threat detection system, which is radar based,
to scan the intersection's highest risk areas during the test vehicle's
approach of the intersection.
Knowing each intersection's "threat regions", a radar based system can optimally scan the intersection for possible conflicts. This system consists of a three-radar assembly that will monitor traffic in the same fashion as the tracking radar systems previously mentioned. But, the intersection area provided by the GIS will then be used to develop the appropriate gap time, which utilizes the time predicted by the tracking radar for vehicles to enter and exit the intersection. A collision-warning algorithm would then obtain the gap time metric and issue warnings to the test vehicle if the gap time is insufficient for a safe crossing, or if it and any other vehicle are predicted to occupy an intersection simultaneously.
This project is a good example of a combination of technologies used to enhance current intersection collision avoidance capabilities. The joining of GPS/GIS information with radardetection is still being evaluated, but results thus far are quite promising. Future endeavors include testing in real-world conditions and use of a multi-beam radar assembly.
As previously discussed, much of the work on Intersection Collision Avoidance stems from the technologies currently being applied to Longitudinal CAS. The added consideration of an additional flow of traffic makes it a more difficult endeavor. The ability to distinguish and successfully manage differing traffic flows and apply this to crash avoidance is crucial as we head toward Intelligent Vehicle Highway Systems (IVHS).
Radar technology has been adequate in recognizing dangerous situations and helping to avert potential traffic conflicts in test situations. And as we have seen in our representative projects, radar technology used in conjunction with other technologies enhance our abilities to identify possible intersection collisions before they happen, with the ultimate goal of giving drivers enough time to make informed decisions and proper judgments before they are involved in a crash scenario.
Jocoy, Edward H., Herbert A. Pirson
"Threat Detection System for Intersection Collision Avoidance - Real -Time System
Performance" Vehicle Navigation Systems and Advanced Controls. Pages 87-98 1999
Najm, Wassim G.
"A Review of IVHS Crash Avoidance Technologies" Collision Avoidance Systems:
Issues and Opportunities Proceedings March 21-22, 1994 Notes from the Collision
Avoidance Workshop
Pierowicz, John
"Development of an In-Vehicle Intersection Collision Countermeasure" Transportation
Sensors and Controls: Collission Avoidance, Traffic Management and ITS Pages
284-290 1996
Author: Bryon Li. Date: 06/26/2000