Compare, if you will, your automobile to the model you might have owned fifty years ago. If you own a new car, it has seat belts and front airbags. In 1949, only those lucky people who bought a Nash-Kelvinator Rambler were treated to the safety of seat belts; if you wanted a Ford or Chrysler with occupant restraints, you would have had to wait until 1955, when those companies introduced seat belts as optional equipment. Your new car also has shatterproof glass, headrests that prevent whiplash, and a collapsible steering column; most of these safety features were as unheard of in 1950 as trunk-mounted CD changers.
Now compare your automobile to one you may be able to buy ten years hence. Analysts predict that 2010 models will have more amenities, such as navigation systems, night-vision enhancement, voice-activated controls, and devices that haven't even been thought of yet. More importantly, future vehicles may also incorporate a new generation of safety devices. Unlike seat belts and airbags, which seek to mitigate the effects of the "second collision" of the passenger striking the inside of the car, these new safety systems may help to prevent crashes by warning drivers of unsafe conditions before a collision occurs or by taking control of the vehicle to avoid a crash.
The Volpe Center is contributing to the development of these Crash Avoidance Systems as a participant in the Intelligent Vehicle Initiative, an innovative partnership between the Federal government and private industry. By working together, the public and private partners hope to develop safe, effective, and cost-efficient technology that will become standard equipment on new vehicles.
Regulating Safety
This new Initiative is the latest in a series of attempts to improve the safety of the automobile. In their book The Struggle for Auto Safety, Jerry Mashaw and David Harfst describe almost a century of efforts intended to protect automobile users as speeds increased and roads became more congested. According to Mashaw and Harfst, the prevailing opinion before 1965 held that motor vehicle collisions, injuries, and fatalities were generally attributed to driver error. Conventional wisdom perceived the vehicle as a neutral interactor and maintained that vehicle safety must be addressed through education and enforcement aimed at behavior modification. Automobile safety features were offered largely at the discretion of auto manufacturers, and when Ford's 1956 safety-oriented marketing campaign fell flat, many industry analysts concluded that safety didn't sell as well as tail fins.
As the annual death toll on the highway rose from 35,000 in 1950 to 49,000 in 1965, motor vehicle safety became a subject of national debate. Mashaw and Harfst describe a different opinion on auto safety that surfaced during the early sixties. This view posited that automobiles, modeled to be stylish and speedy, were dangerous by design. Improvements in the physical safety of autos would provide a much greater safety return than continued efforts in behavior modification.
Crash Avoidance Systems currently being researched at the Volpe Center, use remote sensors (such as radar) to detect the presence of other vehicles or objects, a computer to evaluate the possibility of a collision, and a "driver interface" to relay that information to the driver
The ensuing debate in Congress resulted in passage of the 1966 Motor Vehicle Safety Act, creating the National Highway Traffic Safety Administration (NHTSA). Since that time, the Department of Transportation and the automobile industry have made numerous attempts to improve the safety of automobiles by standardizing safety features or recalling vehicles with defects. During the period 1966 to 1994, the rate of automobile fatalities dropped from 5.5 to 1.7 deaths per 100 million vehicle miles traveled, an improvement of almost 70%. In spite of these improvements, the private automobile remains one of the most dangerous ways to travel. In 1994, the most recent year for which comprehensive data is available, there were 40,676 motor vehicle deaths on America's roadways, 5.2 million non-fatal injuries, and 27 million damaged vehicles. A 1996 report estimated that these collisions cost American drivers, businesses, taxpayers, and insurance companies almost 150 billion dollars each year. This enormous sum is the result of all costs associated with auto collisions, including both direct expenses (property damage, medical costs, legal costs, emergency services costs, insurance administration costs, and premature funeral costs) and indirect losses (travel delays, lost wages, and productivity losses).
In 1991, Congress recognized the need for continued advancements in transportation and passed the Intermodal Surface Transportation Efficiency Act (ISTEA). This important legislation calls for safer and more efficient use of the nation's transportation infrastructure (e.g., highways, roads, transit facilities). Title IV of ISTEA established the Intelligent Transportation Systems Program to apply state-of-the-art technology to vehicles and transportation infrastructure. The program is structured to address four key principles: promote the development of comprehensive Intelligent Transportation Systems (ITS), support research and development, ensure that Intelligent Transportation Systems technology is safe and cost-effective, and emphasize private sector involvement.
Since its inception, the program has pursued advancements in many different areas and has contributed to many projects around the country, including traffic signal optimization, in-vehicle navigation systems, electronic fare and toll payment, and electronic truck safety inspections. ITS is also the nexus of research relating to as-yet unrealized projects such as central traffic management systems.
A key component of the ITS program is the Intelligent Vehicle Initiative (IVI), which promotes the development and commercialization of advanced driver assistance technologies designed to further safety, mobility, efficiency, and environmental quality. The central goal of IVI, safety, is being pursued through the development of advanced systems that may actually help drivers avoid a motor vehicle collision. These Crash Avoidance Systems (CAS), currently being researched at the Volpe Center, use remote sensors (such as radar) to detect the presence of other vehicles or objects, a computer to evaluate the possibility of a collision, and a "driver interface" to relay that information to the driver.
"it is a bold proposition that automotive technology would complement the functional space previously occupied only by human perceptions and cognitive capability"
This preventative technology is in many ways an unprecedented approach to vehicle safety. The systems under development do not passively prevent potential collisions through recalls, or mitigate the effects of collisions through safety features such as air bagsthey actively assess the driving environment in ways that the driver cannot (using remote sensing and computer algorithms) and convey that "potential collision" information to the driver who can then act on it before the collision is inevitable. The Business Plan for the IVI recognizes the fundamental shift inherent in this active prevention: "No aspect of automotive technology has ever tried to accomplish what the human driver does with his or her eyes in terms of assessing the immediate need for speed and path control ... it is a bold proposition that automotive technology would complement the functional space previously occupied only by human perceptions and cognitive capability." This approach emphasizes the significant role of the driver in achieving improved highway safety.
The Intelligent Vehicle Initiative is also important because it operates as a joint partnership between public and private entities, satisfying a key goal of the broader ITS program. This partnership, still in its formative stages, represents a new working relationship in which Federal government and private industry collaborate to promote the development, testing, and deployment of effective and practical safety features. The Federal partner, DOT, takes the lead in early stages of the Initiative, conducting basic research, defining performance standards, and developing objective evaluation methodologies. The Department will also evaluate promising IVI systems via field operational tests and assess the safety benefits of these systems. Private industry (automobile producers, component manufacturers, fleet owners, etc.) will develop the technology, integrate it into the vehicle, conduct field tests, and put it into production.
A New Approach: The Intelligent Vehicle Initiative
The Department of Transportation has also recognized that success of the joint partnership depends on agreement between the goals of the public and private partners. With that in mind, DOT has sought to develop research goals, objectives, and priorities in collaboration with private industry. Recognizing the need for balance between public benefit and private incentive, the IVI Business Plan emphasizes that reasonably achievable public safety benefits must be identified and quantified for each system under development. It states that systems should be commercially viable and amenable to near-term implementation; they should incorporate standardized interfaces and protocols; and they should be acceptable to drivers. By utilizing shared resources and common goals, the joint partnership can overcome many of the hurdles that have impeded Federal auto safety efforts. The result is more effective technology, accelerated development, and well-defined benefits.
Research into Crash Avoidance Systems has been divided into four work areas, or platforms: light vehicles (cars, vans, and light trucks), transit vehicles (buses), commercial vehicles (heavy trucks), and specialty vehicles (snowplows, ambulances, etc.). Each of these platforms has distinctly different dynamics and different needs regarding Crash Avoidance Systems. By focusing on issues specific to each platform and the technology applicable to those needs, the IVI can accelerate development and deployment of systems that will have the greatest benefit to each platform. The Business Plan also outlines "crosscutting" activities that will facilitate work across platforms. These activities include: architecture and standards development; human factors research, development, and testing; acquisition, expansion and validation of evaluation tools such as models; development and execution of an outreach plan to ensure joint participation of industry and other stakeholders; development and implementation of field operation evaluation plans; and program planning and administration.
Volpe Center Support
Recognizing that the development of effective Crash Avoidance Systems was contingent on a rigorous understanding of the collisions themselves, NHTSA asked the Volpe Center to analyze a national database of automobile crash data to analyze crash scenarios and identify potentially effective countermeasures. The 1995 Synthesis Report that resulted from this work identified seven major crash types and specified three crash types that had the greatest potential for prevention through Crash Avoidance Systems: rear-end collisions, lane change/merge collisions, and single vehicle road departure collisions (those in which a vehicle leaves the roadway without striking another vehicle). These three types of collisions account for approximately 48% of all automobile collisions and 36% of all auto-related fatalities. These findings were incorporated into the IVI Business Plan, and rear-end, lane change/merge, and single vehicle road departure crashes have become the three focus areas for Crash Avoidance System research for the light vehicle platform.
One group of researchers at the Volpe Center has been supporting the development and evaluation of Crash Avoidance Systems designed to prevent rear-end collisions. The Crash Avoidance Research Team analyzes crashes, develops guidelines for evaluation of systems, estimates the benefits of prototype systems, and conducts independent evaluations of field tests. The team has developed a "precrash scenario analysis model" that can be used to estimate the effectiveness of prototype systems in different scenarios.
Other Volpe Center staff are actively involved in the development of Crash Avoidance Systems for the transit platform, where collision dynamics and technology needs are distinctly different from those of light vehicles. This group is also compiling baseline statistics that identify the most important Crash Avoidance System needs and provide a benchmark against which prototype systems can be evaluated. This group also organized a forum and roundtable to gather input from the transit industry.
Looking Forward to Rear-End Collisions
By far, the most promising light vehicle crash avoidance research is in the area of rear-end collisions. During the years 1992-1996, rear-end collisions accounted for roughly 1.6 million police-reported crashes annually, constituting roughly one-quarter of all such crashes. Perhaps up to 2 million minor rear-end collisions occur annually but are not reported to police. The prevalence of these collisions may be attributed to many factors, such as poor road conditions, excessive speed, and poor road alignment. However, driver inattention, following too closely, external distraction, and poor judgment are the primary cause of more than 80% of all rear-end collisions. Incidents attributable to these errors could be prevented by a system that would alert drivers when they are in a potentially dangerous situation before it is too late to make avoidance maneuvers.
The rear-end Crash Avoidance Systems currently under development operate by using radar, lasers, or other remote sensors to determine the distance to and relative speed of the leading vehicle. A computer within the car performs calculations to determine if sufficient stopping distance and time are available. The system can alert the driver through visual, tactile, and auditory warnings, such as a voice that says "Look ahead" or "Brake." Warnings may become increasingly more urgent as the risk of collision increases. Future generations of Crash Avoidance Systems may also incorporate the ability to initiate evasive actions, such as braking that may prevent an accident even if the driver does not respond to warnings. It is important to note that the systems under development work autonomously; i.e., without any special equipment in other cars or on the road.
The Crash Avoidance Research Team studying these new systems is part of the Volpe Center's Accident Prevention Division. Team members Wassim Najm, Marco DaSilva, Andy Lam, and Joseph Koziol are engineers who now focus on the development of Crash Avoidance Systems; John Smith, an operations research analyst, provides support to their research. The team seeks to understand how many different factors interact in the moments before a collision. By reconstructing "dynamically distinct rear-end precrash scenarios," the team can determine whether a particular Crash Avoidance System would have recognized that scenario as a potential collision and whether it could have alerted the driver in time to prevent the crash. This information can be used to objectively evaluate different Crash Avoidance Systems and to estimate their safety benefits.
In order to study the factors that contribute to rear-end collisions, Wassim Najm turned to the NHTSA General Estimates System (GES) database, a nationally representative sample of police collision reports. The database contains three main files: Accident File, Vehicle/Driver File, and Person File. These files contain data about, respectively, road conditions and time of accident, vehicle speed and driver characteristics, and age and injury. From these files, Wassim can also extract information about driver avoidance maneuvers, vehicle type, injury severity, and other aspects of collision dynamics. An analysis of 60,000 collisions described in the GES identified the twenty most common precrash scenarios, of which the top five comprise almost 90% of all rear-end crashes.
So what do these precrash scenarios look like? Approximately 37% of all rear-end collisions occur when two vehicles are traveling at a constant speed and the lead vehicle decelerates. Only slightly less common (33%) is the scenario in which one vehicle encounters another vehicle stopped in a travel lane ahead. In 14% of rear-end collisions, both vehicles were in the same lane but the leading vehicle was traveling slower. Roughly 5% of these collisions occur when two vehicles are both decelerating and the lead vehicle decelerates at a higher rate.
The GES database suggests that the majority of the drivers who were involved in rear-end collisions performed no action to avoid the crash. Those who did attempt to steer or brake to avoid the collision constitute only a fraction of those involved. It is important to note that these figures are for drivers who were involved in a rear-end collision. Many other drivers may have been involved in potential crash scenarios but responded in time to initiate avoidance maneuvers and sucessfully prevent the collision. The important role of driver inattention is also highlighted by the finding that a significant majority of rear-end collisions occur below the posted speed limit, either in congested traffic conditions or near intersections. Furthermore, approximately 94% of police-reported rear-end crashes occur on straight roads, suggesting that visibility problems or curves are not to blame. These data suggest that the causes for many rear end collisions lie not outside the car on wet roadways or around blind curves, but inside the car with drivers increasingly distracted by cellular phones, shaving, radios, and a multitude of comfort devices. Those drivers who are attentive may be able to initiate evasive maneuvers in time to avoid the crash.
The Volpe Center is producing a precrash scenario identification process that will allow other researchers to identify dynamically distinct precrash scenarios from future GES data as it becomes available. This identification of precrash scenarios is vital to the IVI. It provides a common base of information that the public and private partners use to develop standards for the collision avoidance systems (it identifies scenarios that every system should be able to recognize), and it provides a basis for estimating the benefits of these systems.
The Research Team will also develop safety benefits assessment methodologies for experimental tests of Crash Avoidance Systems in driving simulators, on test tracks, and in real traffic situations. This methodology will incorporate system usage rates, vehicle miles traveled, driver type, traffic condition, and roadway class so that findings can be extrapolated to different levels of "market penetration." The Volpe Center will also be developing a model to evaluate specific precrash scenarios using statistical techniques such as the Monte Carlo method. This technique is used to analyze systems with large numbers of variables; the analysis provides an overall evaluation of system performance by running thousands of computer simulations based on independently randomized variables. This probability model will be calibrated by comparison to the rear-end crash possibility data from field operational tests such as those of the Intelligent Cruise Control (ICC) System. The Team will also write computer programs that calculate the probability of a collision in different precrash scenarios and estimate safety benefits for different avoidance systems.
Soon the team will have an opportunity to apply its evaluation methodology an upcoming field test. The USDOT in cooperation with General Motors will sponsor an "operational field test" of state-of-the-art rear-end Crash Avoidance Systems. The University of Michigan Transportation Research Institute will conduct the field test and collect data. As the Independent Evaluator of the test, the Volpe Center will review this data, evaluate the effectiveness, and provide an estimation of its safety benefits.
One prototype Crash Avoidance System has already been subject to both controlled on-road studies and driving simulator tests in the laboratory. Wassim Najm compared the results of these tests to the GES database to determine how many rear-end collisions the system could have prevented. His findings indicate that, if 100% of all cars were equipped with the system, and if all drivers used the system properly, almost half (47.7%) of the rear end crashes could have been prevented, potentially saving 10.7 billion dollars annually. Furthermore, one related system has also been tested in real-life driving conditions. Field tests were conducted on an Intelligent Cruise Control system that uses remote sensors to measure the gap to the leading vehicle and adjusts the throttle of the car accordingly. The Volpe Center's evaluation of the field test determined that the system has the potential to prevent approximately 12,000 rear-end collisions on interstate highways, based on 1996 GES crash statistics.
Safe Buses Get Safer
Led by the Federal Transit Administration, the DOT has also been investigating ways to apply intelligent vehicle technology to transit vehicles such as buses and vans. Compared to personal autos, mass transit is a remarkably safe way to travel. Regardless, there remains room for improvement. The 24,000 bus crashes that occur annually account for 42,000 injuries and $800 million in insurance claims.
Nora Burke, a researcher in the Infrastructure Protection and Operation Division, has been part of the effort to develop Crash Avoidance Systems for transit vehicles. She compiled data on transit collisions and presented it at two forums, organized by the Volpe Center, where members of the transit industry (transit authorities, manufacturers, component suppliers, researchers, planning agencies, etc.) met to discuss the development and implementation of new technologies. The results of these two forums were published in two separate reports entitled Transit Intelligent Vehicle Initiative (IVI) Forum and Roundtable. While crash avoidance technologies were identified as the most important need for transit vehicles, future transit technologies may address precision docking, security, customer information, and vehicle diagnostics and servicing. Attendees also agreed with the analysis provided by the Volpe Center's Crash Avoidance Research Team that rear-end and lane-change/merge collisions are of major concern to the industry.
These two types of collisions are central to transit Crash Avoidance System research. The roughly 5000 rear-end crashes that occur annually constitute approximately 21% of the total collisions in which transit vehicles are involved. Buses are primarily involved in rear-end collisions when they are struck from behind by another vehicle. Technologies that may prevent these crashes include warning systems, such as high-visibility lights or flashers, mounted on the rear of the bus and designed to alert the driver of vehicles in the rear. Other collision mitigation systems may alert passengers within the bus of an impending collision. Lane-change/merge crashes happen slightly more often than rear-end collisions; approximately 8640 such collisions occur every year, more than one-third of the total. These collisions are are a problem for transit vehicles because these vehicles make frequent wayside stops and travel lane re-entries. Furthermore, the physical configuration of transit vehicles results in limited visibility and a larger blind spot than personal autos.
Nora Burke's team is currently compiling this crash frequency data in a Baseline Statistics Report that will also include information on severity of crashes, associated injuries, and the movement and corrective actions taken prior to the collision. Her preliminary findings indicate that, although lane change/merge collisions are the most common crash type for transit vehicles, they are generally not very severe, and commonly involve minor property damage for all vehicles involved. On the other hand, vehicles who strike buses in rear end collisions almost never perform corrective action; because they hit the bus at higher speeds, this type of collision is often severe and may involve both property damage and injuries to bus passengers. This and other data compiled in the baseline statistics report will facilitate the prioritization of Crash Avoidance System applications and will provide a benchmark against which the performance of crash avoidance systems can be evaluated.
Once avoidance technologies for transit applications are developed, the industry will be in a unique position to implement the new systems. Unlike passenger autos, transit vehicles are operated and maintained by well-trained professional workforce that can adapt to new systems and ensure their effectiveness. The existing industry institutions, infrastructure, and standards can also be utilized to implement new technology. While questions remain regarding training, liability, and labor issues, the transit IVI platform benefits from strong industry support. This support and participation will be crucial to developing systems that are feasible and effective.
The next step for the transit platform is to develop performance specifications for Crash Avoidance Systems. The Federal Transit Administration has funded three performance specification projects. The performance specifications will detail requirements for Crash Avoidance Systems that will address safety issues related to lane change/merge and rear-end collisions.
A Successful Partnership
Wassim Najm is in a uniquely qualified position to participate in the IVI program, since he has already been the Contract Officer/Technical Representative for a successful Joint Partnership that accelerated the development of compact "night vision" systems. That $25 million partnership, initiated in June 1995, involved equal cost sharing between the Defense Advanced Research Projects Agency (DARPA) and a technology industry consortium that included Texas Instruments (now Raytheon Systems), Delco Electronics, and Marlow Industries. The program was a part of the Technology Reinvestment Project, an initiative designed to commercialize defense technology. The Project results in both "spin-off" and "spin-on" benefitsthe private partners apply advanced defense technologies to commercial applications, while the public partners benefit from more cost-effective production techniques developed by private industry.
Prior to the joint partnership, Texas Instruments had researched infrared night-vision systems for nearly ten years but had not developed a system compact enough to be feasible for use in automobiles. By sharing resources and information, the public-private partnership accelerated the development of low-cost night vision systems by 3-4 years and reduced the anticipated overall system cost by a factor of three. A Technical Advisory Board, comprised of potential night vision customers such as the U.S. Coast Guard, the Federal Bureau of Investigation, the Army, and law enforcement agencies, provided input critical to the program.
The night-vision systems have been so successful and cost-effective that General Motors plans to incorporate infrared night-vision displays on the 2000 Cadillac Deville. The infrared image will be displayed via a small projector that projects the image onto the windshield approximately 3 degrees below the driver's line of sight.
References
Blincoe, Lawrence J. The Economic Costs of Motor Vehicle Crashes, 1994. National Highway Traffic Safety Administration, Washington, D.C., 1996.
Implementation of the National Intelligent Transportation System Program, 1996 Report to Congress. Joint Program Office for Intelligent Transportation Systems. U.S. Department of Transportation, Washington, D.C., 1997.
Koziol, Joseph S. and Vaughan W. Inman. Safety Evaluation Methodology for the Intelligent Cruise Control Field Operation Test. pp. 287-298, in Object Detection, Collision Warning and Avoidance Systems. Automotive Electronics Series, PT-70. Jurgen, Ronald K., Ed. Society of Automotive Engineers, Inc., Warrendale, PA, 1998.
Mashaw, Jerry L., and David L. Harfst. The Struggle for Auto Safety. Cambridge: Harvard University Press, 1990.
Najm, Wassim, Christopher J., Wiacek, and August L. Burgett. Identification of Pre-crash Scenarios for Estimating the Safety Benefits of Rear-End Crash Avoidance Systems. 5th World Congress on Intelligent Transportation Systems. Seoul, Korea, October 1998.
Najm, Wassim, Mark Mironer, Joseph Koziol, Jr., Jing-Shiarn Wang, and Ronald R. Knipling. Synthesis Report: Examination of Target Vehicular Crashes and Potential ITS Countermeasures. John A. Volpe National Transportation Systems Center.
Resendes, Raymond J. The Intelligent Vehicle Initiative: A Report. ITS Quarterly, Summer 1998.
Wiacek, Christopher J., and Wassim Najm. Driver/Vehicle Characteristics in Rear End Precrash Scenarios Based on the General Estimates System (GES). Society of Automotive Engineers, 1999.
