This article also appears in
Subscribe now »

Stereo-camera vision developed in the UofM-Dearborn Visual Odometry lab, along with lidar sensors under study, will define the location and trajectory of  objects in the shuttle’s path. The development team aims to improve the route infrastructure by programming traffic signs and overhead lights to communicate with the vehicles. (U of M-Dearborn Electrical and Computer Engineering Dept. illustration)

U of M-Dearborn students engineer new autonomous shuttle system

Evolving to stay relevant in the fast-changing Engineering world is a challenge faced by every institution of higher learning. In 2013, educators at the University of Michigan’s Dearborn campus inspired by a National Academy of Engineering study took this evolution seriously with an initiative called “Preparing the Engineer of 2020.”

Outdated lectures, text books and lab experiments were replaced by a fresh Design, Build, and Test curriculum. New lean-startup courses addressing customer discovery techniques were added. The goal was to assure that engineering students graduated with the skills they’d need to be productive at the very start of their careers.

A prime example of the Design, Build, Test approach is an autonomous shuttle under development for implementation at the Dearborn campus next year. The seed for this project was planted when the nearby Henry Ford Museum embraced autonomy as one of its core areas of emphasis. Why not provide an autonomous shuttle to ferry visitors between the Amtrak station and the museum, one U of M-Dearborn educator wondered. After a few months of study revealed the broad scope of such an undertaking, the idea evolved into a more attainable on-campus service dedicated to faculty, staff and student use.

Anthony England, dean of the College of Engineering and Computer Science, recognized the merits of the project and authorized seed funding earlier this year. Engineering professors and their students then began building the shuttle’s foundation: selecting a suitable vehicle platform; engineering drive-by-wire systems for guidance, propulsion, braking, and signaling; and creating three-dimensional GPS maps of the campus.

The vehicle of choice is a six-passenger Polaris GEM. This light-weight "neighborhood" EV has a top speed of 25 mph (40.2 kph) and quick battery-charging capability. All the controls and equipment necessary for by-wire operation have been designed, built and tested in the laboratory per the university's new educational doctrine. GPS maps now in hand will enable the first test vehicle to drive autonomously on the planned 1.8-mi (2.8-km)-long campus route sometime next year. By the end of 2018, free shuttles are expected to be in full service.

Electrical and Computer Engineering Chairman Paul Richardson considers the Dearborn campus an ideal environment because it will expose these autonomous vehicles to real world societal influences—such as unpredictable student pedestrian behavior—without requiring compliance with Michigan DOT strictures—such as the need for human drivers.

The next phase is fund raising—securing partners willing to share some of the cost in exchange for access to the autonomous shuttles and their operating environment. Industry partners currently are supporting 60% of the research conducted at U of M- Dearborn.

Explained Sridhar Lakshmanan, an Electrical and Computer Engineering professor, "Standards for the performance of lidar and other sensors don’t exist today. Our live autonomous laboratory provides an excellent means of filling that void.” He adds that autonomous capability will be phased in incrementally.

“We’ll start with the GPS map base which identifies roadways, traffic signs, and the like but not variable hazards such as pedestrians," he noted. "Then we’ll add stereo-camera vision developed in our Visual Odometry lab to define the location and trajectory of any object in the shuttle’s path. Lidar sensors—also under study here—will add more size, distance, and object movement detail."

The team aims to improve the infrastructure of the shuttle route by programming traffic signs and overhead lights to communicate with the electric people movers. The final environment will be safe, convenient, and highly functional, he asserted, "while providing industry partners with an excellent means of testing their sensors and equipment well in advance of their production applications.”

Lakshmanan and Richardson are quick to point out the key difference between the Dearborn shuttle and a similar service that begins later this year at the University of Michigan’s Ann Arbor campus.

“The French supplier NAVYA is manufacturing the Ann Arbor shuttles in ready-to-operate form,” noted Lakshmanan. “While that system will be useful for gathering autonomous vehicle data, engineering students there will miss out on the design, build, and test opportunities we’re offering with our Dearborn project.”

Continue reading »