Safe Plowing - Applying Intelligent Vehicle Technology
Improving highway safety and increasing mobility are two of the bedrock goals of the Federal Highway Administration (FHWA). These goals are constant - 24 hours every day, seven days every week - in good weather and bad.
In treacherous winter weather, FHWA meets some of its toughest challenges to safety and mobility. In some states, such as Minnesota, weather is a contributing factor in one out of every five vehicle crashes.
When weather forecasters predict snow or blizzard conditions, they often warn people to stay at home and to keep off the roads. Usually, they will mention the threat of slippery road conditions that can result in a rash of car crashes. However, there is less discussion of another serious winter hazard - the increase in cross-lane and run-off-the-road crashes because the road markings and boundaries are obscured by snow.
We count on our highway maintenance people to clear the snow off the roads, but how do the snowplows stay on the road? How can the snowplow operator find the edge of the road when it is buried under deep snow? This can be a problem anywhere, but it is particularly dangerous in mountainous terrain where the consequences of running off the road are most severe.
In November 1999, former U.S. Secretary of Transportation Rodney E. Slater announced that a $3.9 million grant was awarded to the Minnesota Department of Transportation (MnDOT) to conduct a Specialty Vehicle Field Operations Test. Combined with the contributions of other partners in this three-year project, the total budget is $6.5 million. This project will tap intelligent vehicle technology to enable snowplow operators to "see" the roadway that is hidden to the human eye by deep snow or white-out conditions and thereby improve safety during plowing operations while also decreasing road closures or slowdowns due to winter weather conditions.
Because snowplow operators routinely face the hazardous duty of clearing slippery, snow-covered or ice-coated roads with perhaps the additional dangers of high winds, blowing snow, or white-out conditions, intelligent vehicle technology is being rigorously tested in Minnesota on state-owned snowplows. In addition to the hazards from the weather, snowplow operators must also cope with the dangers of hidden objects covered by snow and with improper visual cues from previous plowing. MnDOT snowplows are involved in 131 collisions annually, while in 1999, the California Department of Transportation (Caltrans) recorded 194 accidents for vehicles involved in snow operations. In 1998, these accidents cost MnDOT $1.8 million in property damage and $450,000 in damage to Twin Cities snowplows, and in 1999, the cost to Caltrans in snow removal operations for vehicle accidents and personnel injuries was $390,679. That's a hefty annual price tag for both states.
But how do you avoid plowing into something you cannot see? Give snowplows 360° radar to detect moving objects, make use of Global Positioning System technology and a geo-spatial database that enables a driver to determine where lane boundaries and the presence of fixed objects such as guardrails and mailboxes, and then alert the driver via the vehicle's collision warning system. Caltrans has successfully demonstrated the use of magnetic guidance technology in which the vehicle is guided by an in-vehicle computer that tracks magnetic markers embedded along the center of a lane 1.2 meters apart to define the roadway.
If the snowplow operator can't see the road, what about the motorist that may be traveling behind the snowplow with his/her visibility decreased by the snow throwing of the plowing operation? Give snowplows a rear-mounted, external strobe light to keep motorists at a safe distance and to reduce rear-end collisions between motorists and snowplows.
Still, there is the battle that snowplow operators fight with stress brought on by the intense concentration needed to maintain safe and efficient control of a snow-removal vehicle. How do you avoid creating more driver stress through information overload? Give drivers a voice during the development and testing period so that their feedback can influence improvements to the design of intelligent vehicle enhancements and test under real conditions using real snowplows and real operators.
Winter weather can lead to another problem - potential damage to the nation's economy. When road closures due to winter weather are unavoidable being able to operate just two hours a day has an incredible impact on the economy. Safe and efficient plowing is an economic requirement. Closed or impassable roads reduce mobility and wreak havoc with deliveries of goods to businesses and with the ability of workers to get to their places of employment.
"In Minnesota, if all major state roads were closed for 24 hours, estimates of the cost in lost wages, lost retail activity, and lost economic activity exceed $100 million," said William Gardner, Intelligent Vehicle Initiative (IVI) program manager for MnDOT's Guidestar Intelligent Transportation Systems (ITS) program.
In California, four of the eight major highways into and out of the state require snowplow operations to keep them open during the winter. Closure of Interstate 80's Donner Pass would affect shipment of goods to Nevada, Idaho, and Utah as well as affecting the 21 ski resorts around Lake Tahoe and the entertainment business in Reno, Nev. The financial effect of a closure of the Donner Summit for 24 hours on gaming interests in Reno alone is estimated at approximately $10 million. I-5. which runs north-south the length of California, is a major transportation corridor for international commerce related to the North American Free Trade Agreement, and a closure of I-5 would have serious international trade consequences. So, better plowing not only means increased safety, but it is also good for business.
High-Tech Solutions to Age-Old Problems
Researchers involved in the Specialty Vehicle Program Partnership are working together on two major efforts headquartered in Minnesota and California. While the goals are the same - develop and test intelligent vehicle-enhanced snowplows that are safer and more efficient - the approaches differ due to regional weather conditions.
"California is more heavily populated than Minnesota and must contend with heavy traffic as well as mountainous conditions with deep snow cover," said Roy Bushey, program manager for the Caltrans New Technology and Research Program.
"Unlike conditions in California, it's not the depth of the snow, but the blowing, that creates problems here," said MnDOT's Gardner.
Drivers in both states must deal with the problem of poor visibility caused by blowing snow. However, Minnesota has more prolonged ice and snow problems, and California has snow to a greater depth. Minnesota generally gets about 1.25 meters (50 inches) of snow per year, and the state spent more than $31 million on snow removal during 1997-1998. California receives an average of 10.7 meters (35 feet) of snow in the mountains of central and northern California, and the state spent more than $28 million for snow removal in 1999-2000.
Plowing - Minnesota Style
The efforts of Minnesota, with partners 3M, the University of Minnesota, Altra Technologies, the Minnesota Department of Public Safety, McLeod County, Hutchinson Ambulance, and FHWA, are aimed at creating "a driver-assistive system that will help these vehicles stay in their lanes and avoid crashes," said Gardner.
"All of the technologies to be employed have been under development for several years and currently are in testing on snowplows on Trunk Highway 19 and Trunk 101," Gardner said.
The Minnesota team's efforts have yielded snowplow enhancements that rely on the use of a differential global positioning system (DGPS) and a geo-spatial database to locate fixed objects, such as lane boundaries and signposts. Carrier Phase DGPS can be accurate to the two-centimeter level. This highly accurate DGPS, combined with highly accurate geo-spatial databases (elements of the database are mapped to accuracies of greater than 15 centimeters), provides a high-fidelity means to provide lane-keeping information to a driver. The geo-spatial database can be constructed from a number of sources, including photogrammetry data and drive-overs by vehicles equipped with highly accurate DGPS and data acquisition equipment. The geo-spatial database is stored in a computer onboard the snowplow.
Collision-avoidance information is sensed by a radar array on the vehicle, and it uses the geo-spatial database to determine which radar returns arise from fixed objects in the geo-spatial landscape that pose no threat to the driver and which returns arise from obstacles that do pose a threat. Only those returns that indicate a threat are given to the driver in the form of a warning through the driver interface. This minimizes false alarms to the driver, and increases the drivers' acceptability.
A Magnetic Lateral Warning and Guidance System developed by 3M uses a special magnetic tape to "outline" the lane. This magnetic pavement marking tape can be used in place of regular lane striping. The tape can be either grooved into the existing pavement and secured with an adhesive or underlayed during construction, and it is detected by a magnetic sensor on the snowplow. The sensor indicates to the driver the vehicle's lateral position within the lane, has a lateral detection range of +/- one meter (approximately three feet), a detection height of 15 to 45 centimeters (6 to 8 inches) referenced from the magnetic tape to the sensor, and an accuracy of +/- two centimeters or +/- five centimeters depending on the lateral distance relationship of the sensor to the magnetic tape.
A central computer interprets the data from subsystems to paint an image of what the road would look like if weather conditions were not preventing the driver from seeing it. This image is projected onto a partially reflective, partially transmissive curved piece of ground optical glass that the driver looks through. Developed by the University of Minnesota, this Heads Up Display (HUD) flips down much like a sun visor so that it can be used when needed and placed out of the way when visibility is not bad enough to warrant its use. Using the HUD, the driver can see the lane boundaries projected onto the snow-covered roadway and can see the location of obstacles that impede safe travel. Looking through the HUD, the driver focuses about nine meters (30 feet) in front of the snowplow, which is normal for most drivers.
"A key aspect is to design these systems so that they are useful and not burdensome to drivers," said Gardner.
"You're not invincible with intelligent vehicle technology," warns John Scharffbillig, the project's technical services manager. "But this is an added safety measure for situations when snowplow drivers would have to be out clearing the roads anyway."
To ensure safety, critical subsystems have backups. Multiple radar devices are used so that if one is not operating, another can take its place and transmit the required data. While asking snowplow operators to trust intelligent vehicle technology, developers are aware that that trust must be earned. System redundancy is helping to gain that trust; nevertheless, Scharffbillig admits, "it takes a certain amount of faith."
One assumption is that real-time DGPS communication will not be available 100 percent of the time when it's needed. That is, the link between the snowplow and the DGPS satellite will fail. Although the latest DGPS receivers reacquire lock in 10 to 15 seconds, this is more than enough time for a plow performing snow-removal operations to go off the road or cause a collision. However, inertial measurement provides guidance during the loss of satellite lock while also providing vehicle-orientation information. Challenges arise as the snowplow moves in and out of signal range, causing communication dropouts. One solution to sustained loss of satellite lock is to determine the lane boundaries by detecting the magnetic tape on the roadway.
System developers use familiar controls to make operators more comfortable and to earn the operators' trust more easily and quickly.
"We use a joystick to control the hydraulics; the HUD is like watching a TV screen; and the controls use eye-thumb coordination like a video game," explains Scharffbillig.
A hallmark of the Minnesota intelligent vehicle development project has been the developers' insistence that all of the intelligent vehicle equipment must be usable without requiring the operator to make an unnatural motion. For example, when using the HUD, drivers need not take their eyes off the road. They perceive the enhanced road image as painted over what they would normally see through the windshield. All of the data is integrated into a single display that requires no head turning and allows the driver to keep his hands on the steering wheel at all times.
California has taken a different approach using a display firmly attached to the cab for safety reasons. In case of an accident, there is nothing that might come in contact with the driver's head.
The California Solution
The Advanced Snowplow Driver Assistance System (ASP) is under development in California and Arizona. This program combines the efforts of the Advanced Highway Maintenance and Construction Technology (AHMCT) Research Center at the University of California at Davis (UCD), the California Partners for Advanced Transit and Highways (PATH) at the University of California at Berkeley (UCB), and the Western Transportation Institute (WTI) of Montana State University (MSU). The California DOT (Caltrans) and the Arizona DOT (ADOT) have provided test sites and snowplow operators.
Now in the third winter of development, ASP, also know as RoadView™ is using intelligent vehicle systems and advanced vehicle control and safety systems (AVCSS) technologies. ASP has been tested through two winters of heavy use in both California and Arizona. Two additional snowplows are being developed for further testing this winter and next winter as part of the RoadView project. Major ASP components include a main computer, a human/machine interface (HMI) with a visual display that shows the snowplow operator how the snowplow is positioned in the lane, and two forward radar sensors that allow full coverage of three lanes ahead of the snowplow - useful when using either a left- or right-mounted wingplow. Azimuth angle data collected from the radar sensors also enable the system to map detected obstacles and inform the driver about the obstacle and the specific lane in which that obstacle is located. The system also provides the distance (rounded to the nearest foot) to the nearest obstacle.
Off the shelf magnetic markers are embedded into the roadway to create a marker reference system. These
magnetic markers can be coded to provide various roadway information by arranging their magnetic poles in specific patterns, which can be read via onboard magnetometers. The California team is currently extending this discrete magnetic marker approach to automate the steering of a 4,000-ton-per-hour rotary plow.
Because any marker at the pavement level or higher might be scraped away by the snowplow, embedded materials were chosen for use in this study. At Donner Summit, the painted lane markings are usually completely scraped off by the end of the winter season, often much more quickly.
Information collected from the magnetic markers and other sources is relayed to a computer that interprets the data and provides an image on the HMI visual display. When an obstacle is sensed, the Collision Warning System (CWS) displays this information on the HMI screen. The HMI is located where a rear-view mirror would traditionally appear and presents information to the driver in a single, coordinated interface.
Developers gave considerable thought to the method of presenting information gathered by intelligent vehicle systems. Early experiments verified the research team's expectation that a "look down" sensing system must provide "look ahead" data so the operator has an indication not only of where the vehicle currently is but where it will be. After all, people drive by looking out at the road, not through the floorboard. To support this, a steering shaft encoder was added to the system to support the prediction of the vehicle's path. Being able to "see" the upcoming curves in the road, the vehicle's current location, and the vehicle's predicted path, the driver can determine the appropriate steering angle. The ability to accurately provide this prediction is a significant advantage of the magnetic marker-based approach.
The California platform provides the driver with a highly accurate view of where the vehicle is and a prediction of where it will be, whereas the Minnesota approach presents an augmented view of the road as it appears in real time to the driver. Different approaches for different circumstances, with both sets of enhancements making plowing safer and more efficient.
What Will It Cost?
The current estimate for a mass-produced California ASP unit is in the range of $20,000 to $30,000, which does not include the vendor's profit margin. Cost of infrastructure installation for the test sites is approximately $11,000 per kilometer ($17,000 per mile), including surveying, installation, and magnets. The infrastructure installation costs have been reduced over the past few years.
"Costs have come down, which creates more potential for new equipment to be cost-effective or even for retrofits to be cost-effective," said Minnesota's Scharffbillig. "The basics of these technologies have been out there. We're just refining them to meet our needs."
On the Horizon
In both Minnesota and California, operators and managers continue to provide valuable feedback, and the research teams are making improvements to the system based on these suggestions. The rapid deployment of the California ASP system into operation - the first plow deployed within five months of the project start in 1998 - represents an early success in the application of intelligent vehicle and AVCSS technologies for specialty vehicles. In the long term, there are no limitations to the application of the ASP technologies because these technologies are applicable across all vehicle platforms, including passenger vehicles, commercial vehicles, and transit vehicles. The same is true of the Minnesota research, either technology may find its way into the family car in the future.
Robert A. Ferlis is the team leader of the Enabling Technologies Team for the Office of Operations Research and Development in the Federal Highway Administration. He has served as the cross-cutting coordinator of the Intelligent Vehicle Initiative (IVI) Program since 1998. In this position, he supports research in vehicle-highway cooperation and is currently responsible for managing the IVI research in advanced snowplow technology. Before joining FHWA in 1997, he worked for 10 years as a transportation research consultant with KPMG Peat Marwick and another 10 years as a systems manager in the energy industry. He received a bachelor's degree in engineering from the University of Illinois and a master's degree in civil engineering from Northwestern University.
Shahed Rowshan was formerly a highway research engineer on the Enabling Technologies Team for the Office of Operations Research and Development in the Federal Highway Administration. He worked in FHWA research, development, and technology from 1990 to 2000, and he was FHWA's IVI specialty vehicle technical director from 1999 until his resignation from FHWA in January 2001. As technical director, he managed the IVI research in advanced snowplow technology. He received a doctorate in civil engineering from the University of Maryland and is a registered professional engineer.
Cathy Frye is the founder of The Fresh Eye, a woman-owned sole proprietorship established in 1994 to provide writing, editing, and publications management services. She holds a degree in writing from Johns Hopkins University and has more than 20 years of experience as a writer and editor. She has worked on various FHWA projects in the past, including the 1997 and 1998 Research and Technology Program Highlights reports. When not writing on transportation issues, Frye can frequently be found working in health care. Her most recent project was serving as the editor of Perioperative Services, a comprehensive resource book for operating room managers. Published last April, this book is already in its second printing.
|Working to Make High-Tech User-Friendly
Developers of driver-assisted systems for snowplows are cautiously addressing human factors issues.
"We're trying to present drivers with additional information without overloading them," says William Gardner, Intelligent Vehicle Initiative program manager for the Guidestar Intelligent Transportation Systems Program in Minnesota. The Minnesota team's improvements are capable of projecting an outline of the roadway - including line markings - onto a flat glass screen that the driver can look through.
Avoiding information overload means listening to the drivers' reactions to the equipment and working closely with them. The project's spirit of cooperation between the snowplow operators and developers is noteworthy, recognizing that even small changes can make a big difference. For example, initially, the Heads Up Display (HUD) used on Minnesota's test snowplows presented all road markings in monochromatic yellow until the drivers requested that white lines be shown as white and the yellow-marked lines as yellow.
The extra information provided by color coding was more useful to drivers than developers had assumed. Today, the system uses color coding.
Also, originally, moving obstacles being tracked were identified by a circle placed around them. The circle would grow larger as the object came closer, smaller as the object receded. Unfortunately, this presented a problem with depth perception. Developers are now testing various icons, including a semicircle and open square, for targeting objects.
Other system elements have also been adjusted to reflect driver input. For example, the vibrational warning in the seat and steering wheel that acts as a virtual rumble strip was annoyingly strong for some drivers, while others reported that it was too weak to get their attention. This feedback is helping the Minnesota Department of Transportation find the most widely accepted setting. Screen luminescence is another system element that varies widely. "This seems to be tied to age," said John Scharffbillig, the project's technical services manager. He noticed that older drivers requested a brighter, more intense image and points out that resolving the problems discovered in many of these observations required the expertise of the University of Minnesota's Human Factors Research Laboratory.
Continuous field testing by the California Department of Transportation's Maintenance Program has yielded the same type of comments from drivers and has resulted in similar system modifications. Initially, their human/machine interface (HMI) display was mounted on top of the dashboard near the driver's right hand. The guiding principle for the display is to provide the necessary information to the operator without any unnecessary informational clutter. When one driver suggested moving the display to the rear-view mirror position, his explanation was simple: give shorter drivers a less obstructed view of the right-hand wingplow mirror. Interviews with other drivers revealed that this was a popular idea, and the design was quickly modified to incorporate it.
While augmenting human senses with intelligent vehicle technology, developers have a heightened awareness of the possibility of introducing what Scharffbillig calls "unintended consequences." The solution is for developers and snowplow drivers to rigorously test the equipment and address the human factors issues raised by incorporating high-tech equipment into snowplows. Through their work together, the resulting improved snowplow is well on the road to being a human-friendly machine designed to give the driver a significant advantage in safety and efficiency over today's equipment.