Wednesday, October 8, 2014

Wheel genuises: UAB team is teaching Army's self-driving trucks a new way to move

Amazon got the world talking with its promise of aerial drone deliveries. Google is making progress toward its dream of a driverless car. But the U.S. Army has already surpassed the tech giants with an operational convoy of robot-driven trucks capable of traveling up to 40 miles per hour.

A self-driving convoy could deliver supplies without putting soldiers in harm’s way, or let those soldiers keep their eyes out for bandits instead of keeping them glued to the road. The Autonomous Mobility Appliquè System (AMAS), built for the Army’s Tank Automotive Research, Development and Engineering Center (TARDEC), has demonstrated its prowess in several online videos (see an example below).

Now researchers in the UAB School of Engineering are working with TARDEC on an even more powerful unmanned system — one that will use smart tires, enhanced sensors and some very quick thinking to guide trucks safely over rough terrain.


Vladimir Vantsevich, Sc.D., Ph.D., professor in the Department of Mechanical Engineering, and director of the UAB Vehicle and Robotics Engineering Laboratory, is an expert on the unique design challenges of multiwheeled vehicles. He has teamed up with UAB Ph.D. candidate Jeremy Gray, who is also a member of TARDEC’s Ground Vehicle Robotics group, on the unmanned convoy project.

Teaching a six-wheeled, 18-ton truck to make smart driving decisions is one problem. String several more behind it, and the challenges multiply. “Imagine: no drivers, just five trucks, following the lead vehicle,” said Vantsevich. One issue is that, even though they are in a line, each vehicle is experiencing different terrain.

“They’re following the same track, so each vehicle will compress the soil a little more, changing its physical properties,” Vantsevich said. “How do you redistribute power between the wheels to overcome this? If one gets stuck, how do you teach the others to avoid that obstacle? No one has ever done this before.”

Big Wheels Keep on Turning

In 2013, Vantsevich and Gray, along with TARDEC’s Jim Overholt, presented an algorithm that unmanned vehicles can use to react to changing ground conditions in real time. Putting that method into practice has required them to make several technical leaps.

Compared to Google’s self-driving car, “an off-road vehicle requires much more information about its surroundings,” Vantsevich explained. A car driving on a highway will pretty much experience the same interaction between tire and asphalt throughout its trip, he says. “But a wheel going over off-road deformable terrain is experiencing continuous changes in its dynamics and motion.”

The UAB researchers are dealing with this challenge by developing tires that read and react to their environment at unprecedented speeds — fast enough to respond to an obstacle while they are moving over it. “You have 60 milliseconds to understand what is going on with the tire, make a decision — should you change the torque of the tire, and in which direction? — and send a signal to the motor controlling that tire,” Vantsevich said.

At One With the Road

To accomplish this, the engineers are designing new types of high-speed sensors, and embedding them in the trucks’ tires and wheels. They are also devising ways to transmit this information from truck to truck, giving following trucks early warning about approaching hazards and terrain conditions.

(Article continues beneath graphic)

Vantsevich declines to describe the new sensors and transmission methods in detail while patents are pending. He is no stranger to innovation, with 30 certified inventions related to the dynamics, energy and fuel efficiency of multiwheel-drive vehicles. In a related project, Vantsevich and Mostafa Salama, a Ph.D. candidate at UAB, are developing new control algorithms to boost fuel efficiency in unmanned vehicles by giving each wheel its own electronic brain.

“I want my vehicle to be able to move from point A to point B with minimum power loss, and to do that I need to minimize the power loss that happens between each tire and the terrain,” said Salama. He has already adapted his original mathematical solution to this problem into a working prototype. Now he is refining that prototype in the Vehicle and Robotics Engineering Laboratory.

These techniques could eventually let unmanned vehicles travel farther, and improve the efficiency of even conventional vehicles, Vantsevich notes. A 40-ton truck, for example, could improve efficiency up to 12 percent with this approach, he says. “That represents a huge fuel savings.”

Lessons From the Frontier

This summer, Vantsevich shared details from his unmanned convoy work with researchers from 15 nations at a NATO Advanced Study Institute (ASI) on “Advanced Autonomous Vehicle Design for Severe Environments” in Coventry, England. The ASI, supported by a NATO grant received by Vantsevich, was arranged and conducted with Coventry University and Sweden’s Royal Institute of Technology. Vantsevich is also the editor of two new series of books that explore hot topics in ground vehicle engineering (Taylor & Francis, CRC Press) and robotics engineering (ASME Press).

UAB students can learn the fundamentals of these new fields in several courses that Vantsevich has developed in the School of Engineering. Although they cover everything from robot design to innovative methods of power distribution, the courses have a unifying theme: mechatronics. This emerging discipline takes an interdisciplinary approach to engineering problems, acknowledging that today’s devices are a complex intermingling of mechanical, electrical and computer systems.

Courses such as Systems Modeling and Controls, which Vantsevich taught to undergraduates in the spring 2014 semester, are all part of a mechatronics track that encompasses classes at the undergraduate and graduate levels, Vantsevich says. “We’re sharing our knowledge with a younger generation, and encourage them to work in these directions.”

No comments:

Post a Comment

We encourage and look forward to your comments on The Mix (and in its related social networking outposts).

Comments will be reviewed before they're posted. Those that are not related to the topic under discussion, promote products or use profanity or abusive language will not be included.