Research Highlights: Transportation Research: A Legacy Continued,
an Interview with Max Donath

Professor James Ryan is well remembered for his, literally, ground-breaking work – dropping cars from a crane on the back lot of Mechanical Engineering in the 1950’s. James “Crash” Ryan was a pioneer who not only developed the retractable seat belt and the event data recorder, commonly known as the “black box,” but also made significant contributions to the crashworthiness of the automobile. Today, Professor Max Donath heads a team of professional researchers that are creating new technologies with the goal of lowering road fatalities.

“It’s been a high priority to reduce such fatalities in Minnesota for a long time. Vehicles and roads are much better designed than before; airbags have made a huge difference; but we’ve done about as much as we can with these and with enforcement and education,” he said. “That’s why we’ve been looking at emerging technologies that can yet make a difference. We focus on the driver – dealing with problems of perception, decision making, and response. We are trying to adapt technology to overcome the limitations of drivers.” The Intelligent Vehicles Laboratory (the IV Lab) , headed by Craig Shankwitz, and the HumanFIRST Program, headed by Michael Manser work together with him developing driver-centered technologies.

Donath began working on transportation research in the mid 1990s. “Our first effort was looking at the problem of drowsy driving. Can we track the motion of a tractor trailer at sufficient accuracy to detect and predict lane departure? Can we determine by its dynamic behavior whether the driver was driving tired or not, and can we then prevent that from happening? That started a whole chain of projects.” The IV Lab has been taking advantage of evolving GPS technology and using it in new ways: “We’ve done things with GPS that no one else has,” said Donath. The inexpensive GPS receivers on the consumer market are for the most part, used for navigation and have an accuracy of plus or minus 10 meters. With a network of correction stations deployed around Minnesota compensating for atmospheric effects, satellite clock errors and wobble, the IV Lab has demonstrated that it is possible to reliably achieve a dynamic accuracies of plus or minus five centimeters. Using this high accuracy GPS, the lab has developed digital mapping systems, capable of accurately recording the location of the road’s lane boundaries.

When the Minnesota Department of Transportation came to Donath with a high priority – snowplow drivers operating in blizzard conditions – the IV Lab had already begun developing the technology for a vision-enhancing assistive system. By adapting that technology, they devised a head-up display (HUD) that allows the driver to “see” the road in a virtual environment inside the cab. The system employs a high accuracy digital map inside a computer on the truck. The high accuracy GPS receiver picks up signals from the satellite to compute the plow’s exact position, as it moves. It takes this information combined with the high accuracy digital map and projects a view of the road from the perspective of the driver on a semi-translucent screen in front of the windshield. Lines show up on the screen drawn by the computer, which lie directly over the lane boundaries. The screen’s semi-translucency allows the driver to not only see the road when visibility is good, but also to operate the vehicle under completely white-out conditions. In addition to the vision-enhancing assistive system, the lab has developed haptic systems; integrating seat vibrators and servomotor driven steering wheels that gently pull the driver’s hand and help the driver stay in the lane. Radar on board the plow monitors what is ahead. These systems also work well with gang plowing, where one plow follows another. In such situations the second plow driver cannot see where he or she is going because snow from the first plow often occludes their view.

Snow plow working in "regular" conditions in Alaska.

Their system has proven so successful that it is now deployed in multiple sites in Alaska and Minnesota. “In order to digitize the needed high accuracy maps and to operate these systems on the road through Thompson Pass in Alaska, we had a GPS correction transmitter airlifted in by helicopter and installed at the Divide station. There are no roads to this transmitter. The winds were so fierce that the first station blew off the mountain so they had to install it a second time. We lost a winter season. The systems at present are being deployed for – “those drivers who have to be out there doing their job under dangerous conditions, when there is a major incentive to deploy this technology,” he said. Of course, the snowplow drivers are very enthusiastic about the system. Snowplows in Polk County are also using the system in the wind-swept Red River Valley of northwestern Minnesota. A mining operation in Alaska is also interested in using this technology on their oversize mining trucks. The IV Lab continues to work on next generation technologies, working to bring the costs down.

Snow plow operator using the driver head-up display (HUD)

Another major project underway is outfitting buses with a HUD, a vibrating seat and haptic steering feedback that will help drivers operate on narrow road shoulders. In Minnesota , buses can travel on the shoulder to bypass congestion. “There is very little room, as little as 6 inches on each side, and if the tires go off the edge and into the mud, the bus may not be able to get back up,” Donath explained. The shoulders are the only available roadway that can be dedicated to buses, without building new roads for them. “The idea is to travel in this lane when the regular lanes are blocked by traffic. When for example, the bus goes through an intersection, the system provides guidance, so that the driver need not slow down to pick up the shoulder on the other side,” he said. The University is receiving $4.3 million to deploy the technology on ten buses that will go into passenger service by the end of 2009. Donath added, “Ultimately, we believe that this same core technology can be used to reduce lane departure crashes, which represent one third of all road fatalities in the country. We are not there yet, and because the technology is so expensive we have focused on applications that can immediately justify the investment. But as we develop and evolve these systems, our goal is to get them deployed on all vehicles in order to reduce the high number of lane departure fatalities, which are often associated with driving tired.”

Although intersections make up only a small part of the U.S. highway network, intersection crashes comprise more than 40 percent of all vehicle crashes nationwide. In rural Minnesota, crash records show that approximately one third of all crashes occur at intersections. Twenty percent of all fatal crashes in Minnesota occur at rural unsignalized intersections. The problem for stopped drivers waiting at the intersection with cars whizzing by, is judging what is a safe enough gap for entering or crossing the traffic stream. “Crashes are usually associated with driver error; the driver looked but did not see, the driver misjudged the gap, or there was an obstructed view,” said Donath. “These are all problems with what is called lag – the distance from the stopped driver to the first approaching vehicle – or gap - the distance between vehicles - selection. How do we help drivers figure out what is a safe enough gap or lag? As you get older you have a harder time judging speed, judging depth. Our focus is on developing the technology to accurately track the gaps and on helping drivers make good decisions about their choice of a gap. This has been a team effort of the IV Lab and the HumanFIRST Program staff. In our experiments at the test site, we use instrumented vehicles to see how drivers scan the field of view when they come to the intersection. Large color animated signs have been installed that convey information in different ways, telling the driver when it is unsafe to enter [the intersection].” Cost is a key issue. The system shouldn’t cost any more than signalizing a rural intersection. “But,” Donath added, “you would prefer not to deploy a signal on a rural road because if you stop the traffic on the main line, not only will you get an increase in rear end crashes, but you will also slow down the traffic on what may be a vital rural corridor.”

Test vehicle at the instrumented intersection.

Teen drivers are another big issue. They represent only 4.7 percent of all licensed drivers nationally, but make up 10.1 percent of all fatal crashes. They represent the greatest risk of any age group. In Minnesota, they make up 7 percent of all licensed drivers, but 14 percent of fatal crashes. Teen drivers are involved in over 18 percent of all fatal crashes, making Minnesota the worst state in the nation, based on data from 2000 to 2006. The causes vary from excessive speeding, to less seat belt use, and more teens driving with other teens in the car. “We have had very poor graduated driving licensing regulations,” Donath said. But in August 2008, the Minnesota legislature enacted new graduated license laws that include a curfew, restricting teens from driving between midnight and five A.M., along with teen passenger restrictions. “We hope it will have an effect. We should see a change in the next few years,” he said.

The IV Lab and the HumanFIRST Program is working on providing feedback to the teen driver so that he or she changes their behavior. We are exploring concepts such as ‘intelligent speed adaptation’. We have developed a working prototype of a cellular phone that includes a GPS receiver, a digital map and a local speed limit database. The phone is mounted on the dash and as a teen drives they get warned if they are driving too fast. If they ignore the warnings and continue to drive over the speed limit, the system text messages their parents, and provides parents with a Google-mapped log of their behavior,” he said. “Besides optimizing how we provide effective feedback, we also need to make sure that the driver can’t turn it off and can’t hack into it. We will also be looking at seatbelt interlocks – the teen won’t be able to drive unless their seatbelt is engaged. There are of course many other issues of concern - privacy, parent/teen relationships, and how to take advantage of these systems in a provisional licensing environment” he added.

A large part of the work is in human factors research; taking the technology and seeing how people interact with it. “We are developing algorithms to trigger the warnings. You don’t want false positives, or to keep warning a driver when there is no reason to warn them,” he said. The displays must inform the driver without being too distracting. “We have an incredible team of technology people and cognitive psychologists who work well together. Once developed and tested, our mission is to get this technology deployed. We are exploring how this can happen, especially now, with the University’s new Office of Technology Commercialization. Everything we do is to try to get this out the door and into the real world.”

The Intelligent Vehicles research team, headed by Craig Shankwitz, also includes researchers, Eddie Arpin, Pi-Ming Cheng, Alec Gorjestani, Arvind Menon, and Bryan Newstrom. The HumanFIRST team is headed by Mike Manser and research staff, Janet Creaser, Peter Easterlund, Justin Graving, and Mick Rakauskas.