Undergraduate Research at Virginia Tech

The Million-Dollar Car

By Ally Haak, English major

Jesse Hurdus and Shawn Kimmel, two Hokies, have begun researching and creating the inconceivable.

How many times have you been driving and wished someone, or something, could take the wheel and drive the car for you? How often have you wished you could communicate with the person in a car near you on the road -- maybe to warn them of something dangerous ahead or to ask directions? Thanks to research teams at Virginia Tech and others around the nation, these luxuries aren’t too far off.

Now graduate students at Virginia Tech, Hurdus and Kimmel began their research as sophomores through the mechanical engineering department. The journey began when Hurdus, intrigued after hearing about volunteer opportunities on an autonomous (self-controlling) vehicle team, introduced himself and Kimmel to Charles Reinholtz, Alumni Distinguished Professor in mechanical engineering and head of the research team.

The students’ interest grew as the possibilities of what autonomous vehicles could accomplish became clearer. Not only could someone use the self-driving capabilities of an automobile for small luxuries, but the creation of such technology could also save thousands of lives. The military has taken a large interest in the development of this project because the use of autonomous vehicles could prevent tragic deaths caused by bombings of supply trucks and convoys. Hurdus says, “Supply missions and other non-combat tasks are the source of an unnecessary number of casualties.”

For the commercial market, autonomous vehicle benefits include a large reduction in vehicular crashes. They allow more precision and quick warnings when departing lanes or coming close to unseen obstacles. Perhaps, in the future, this could lead to a reduction in accidents caused by intoxicated drivers.

Ideas were bountiful and by their junior year both students knew they wanted to advance their volunteer work into undergraduate research.

Futuristic as the possibilities may seem, the technology is surprisingly familiar. Several components lead to the conversion of a commercial vehicle into an unmanned car. This autonomous vehicle must first be converted to “drive by wire.” In other words, the vehicle can be controlled electronically by a computer through a system that the Virginia Tech team designed and used. Hurdus, Kimmel, and others on the research team developed computer software that would be able to control the car intelligently.

It’s not quite as simple as that sounds, however. Running these electronic cars means mastering numerous sensors, actuators, and control. Sensors like GPS and accelerometers are used to determine the absolute position of the car on earth. At the same time, other technologies, including Laser Range Finders and video cameras, actualize the speed and position of the vehicle as well as the position of obstacles and lane and road markings in relation to the car’s movement. All of the sensed data from these technologies are then considered in the creation of control algorithms, which must be used to establish the safe and accurate navigation of the automobile.

Kimmel and Hurdus both worked with software creation and implementation of the “drive-by-wire” system. Later into the project, Kimmel began working on physical systems of the vehicle and actually designed and installed an “electro-hydraulic brake-by-wire system” similar to those found in luxury cars. The work was challenging yet thrilling, he says.

The first car they worked on was a 2004 Cadillac SRX that had been donated to the team by General Motors. “The project itself required a huge amount of testing, tuning, and optimizing, so a large amount of our time was spent in the vehicle,” says Hurdus. “Tuning the speed control for the SRX was one of the most challenging and exciting parts.  The very first time I tested my speed control algorithm, it was like trying to teach an 8-year-old with a lead foot.”

After many hours of tuning, Hurdus was able to “teach” the car to achieve and maintain a speed smoothly and efficiently. “Doing this was a great hands-on experience, applying principles and theories I had learned about in my controls engineering class,” he says. 

The majority of testing was done in a field near the Unmanned Systems Laboratory. While the vehicle was generally a slow learner, its performance continually improved with time. “It was also interesting because if for some reason there was too much lag in the system, or if we tried to drive too fast, the car would act like it was driving drunk," Hurdus says. “This was essentially due to the slower reaction time and we would subsequently try to further optimize and streamline the code.”

The engineers’ hard work literally paid off when, in 2006, the Virginia Tech unmanned systems group received a one million dollar contract to enter a vehicle in the 2006 DARPA (Defense Advanced Research Projects Agency) Urban Challenge. Competition includes such notables as MIT, Stanford, Cal Tech, CMU, and Cornell. Showing further faith in the Virginia Tech team, Ford donated two Hybrid Escapes: one for competition and one for research. And Caterpillar Inc. also donated $100,000 and is interested in other research to automate of some of their trucks.

Ford donated two 2006 Hybrid Escapes. "Being SUVs, they have been nice because they have left a lot of room in the trunk for installing computers and electronics, and the roof-racks are nice for building sensor mounts," says Hurdus.

Needless to say, the undergraduates learned an abundance of information on automotives and autonomous vehicle theory, as well as software development, valuable engineering practicality, and the pressure of working in group situations under strict deadlines.

Both Hurdus and Kimmel agree that the experience was everything they expected and then some. A group of 20-somethings designing and creating a self-driven car sounds reminiscent of a science-fiction novel yet the challenging work of creating, installing, and building software, cars, internal systems, and other technologies should not be taken lightly. Bottom line, as Hurdus excitedly remarks, the team learned “How to turn a full-size commercial vehicle into a giant remote-controlled toy.” The success they have seen and promise they have shown will propel them far into their careers. Kimmel and Hurdus will continue their work as part of their graduate studies and surely will include their research and knowledge in their careers.

For more information see the research team’s website at www.victortango.com. Photos and a video are under the "media" section.