Autonomous Transit Bus Nears Its First Year of Operation

Michigan State University’s SAE Level 4 bus racks up miles in daily use.

Since May 2022, Michigan State University students, faculty and staff have been able to ride within campus boundaries on an autonomous electric bus, one of the largest automated-driving transit buses in the U.S. The bus, manufactured by Karsan, is known as e-ATAK. It features control software by Adastec Corp., which claims to be the first and only company to deploy SAE Level 4 automated buses on public roads, according to Dr. Ali Peker, Adastec CEO.

Turkish bus manufacturer Karsan donated the electric bus that was equipped with autonomous technology, including Adastec’s flowride.ai software platform. (Adastec)

On a 3.3-mile (5.3-km) route, riders can shuttle between the MSU Auditorium and a university commuter lot during daytime hours. With a top speed of 25 mph (40 km/h), the 22-seat, 27-ft-long (8.2-m) automated passenger bus is guided by a sophisticated sensor suite that includes five lidars, eight red-, green-, blue-wavelength cameras, two thermal cameras, one radar sensor, a high-precision global navigation satellite system, 16 ultrasonic sensors, a vehicle-to-everything (V2X) onboard communications module and onboard cellular-V2X onboard communications.

Adastec Corp’s Dr. Ali Peker said his company provided the technology (flowride.ai software platform) to make the e-Atak bus capable of SAE Level 4 operation. (Adastec)
MSU’s John Verboncoeur said bus passengers can complete a survey about their autonomous bus ride experience, to support social science research. (MSU)

The collective array of sensors, cameras and Adastec’s automated software stack (flowride.ai) enable the bus to detect objects, brake, accelerate, change lanes and perform conventional driving maneuvers. Seven intersections along the route are equipped with vehicle-to-infrastructure (V2I) technology to facilitate bus guidance. The e-ATAK bus calculates object movement a minimum of 10 times per second, providing vital information at the intersections, designated bus stops and pedestrian crossings. If a problem occurs, a solution follows.

“We experienced a miscommunication with the technology, which triggered Adastec to develop a system that can override wrong or missing V2I communication signals,” Dr. Peker explained, noting a human operator is ready to take onboard control if necessary.

Prior to its public usage launch at MSU, the bus, its route and the infrastructure were validated by the NHTSA. Testing and validation activities in the months leading to MSU’s automated bus deployment included simulations, x-in-the-loop testing, scenario tests, roughly 630 route test trips and approximately 1,864 mi (3,000 km) of vehicle testing on closed test tracks. Other Adastec autonomous bus deployments are in Stavanger, Norway (a revenue operation that’s fully transit-system integrated), Ploiesti, Romania (at a technology campus), Ankara, Turkey (at the Presidential Complex), Chateauroux, France (one bus), and Germany (a roadshow vehicle), according to Dr. Peker.

The autonomous bus produces relevant data for its development parties. “One of the main lessons we have learned is that even though our software has proven successful in all our deployments around the globe, we still see a lack of human acceptance of riding an autonomous public transit bus,” Dr. Peker noted. That fact has prompted more outreach to educate the public about automated-technology applications.

Raw sensor data and V2X messages are shared with professors for use in classes and for research projects, according to John Verboncoeur, MSU’s senior associate dean for research and graduate studies, College of Engineering; Professor of Electrical and Computer Engineering and Professor of Computational Mathematics, Science and Engineering.

Verboncoeur points out that it often is difficult for researchers and developers to find real-world data from different kinds of sensors. However, MSU students and researchers have access to data collected from the 30-plus sensors mounted on the autonomous bus that has traveled a public road for four seasons. That data provides a framework for developing and testing algorithms.

“Pilot and research projects like this increase the accessibility of data to researchers and students, which gives an opportunity for new research and projects,” Verboncoeur noted.