The Automotive Research Center at the University of Michigan’s Annual Program Review: The Driving Force Behind Autonomous Vehicles
by James Garay, ARC
“In the fall of 1994, the U.S. Army Tank Automotive Command (TACOM) awarded MEAM [Mechanical Engineering and Applied Mechanics] a three-year, $7.5-million research grant to establish an Automotive Research Center (ARC) at the U-M.”1
Twenty-one years later, the ARC, which operates as a consortium of six universities collaborating with partners from both industrial and government sectors, gathered once again on the University of Michigan’s North Campus for its 21st Annual Program Review. Professionals representing each of the three aforementioned domains (academia, government, industry) comprised the event’s nearly 300 attendees.
This year’s event was concentrated around the theme of shared-control and autonomous vehicles, building off of the main emphasis of the ARC, which lies in the simulation and modeling of ground vehicles.
|The opening case study, “No Driver? No Problem: Mobility Across the Autonomy Spectrum in Unmanned Ground Vehicles,” is presented. Click to enlarge.|
The Annual Program Review brings the ARC’s research community together with Army researchers and engineers from industry. This contingent works side by side to review the year’s work, deliberate on the timeline of ongoing projects, and give a weather eye to tomorrow’s automotive challenges. The event’s atmosphere is a dynamic balance between a network of friendships and a professional exchange of information and ideas. It carries the air of a family reunion in bringing all of the ARC’s university partners together, but remains focused on the goal of sharing the latest research developments.
In any case, the environment was one of excitement, as the University of Michigan’s 234- seat Chesebrough Auditorium was packed to standing room only capacity by the conclusion of the opening remarks.
|ARC Director Dr. Anna Stefanopoulou looks on during the poster viewing session. Click to enlarge.|
Dr. David Gorsich, Chief Scientist at TARDEC3, kicked off the presentations with his discussion titled, “The Future of Army Mobility,” wherein he explained the brutal, rugged realities that Army vehicles must be able to conquer. While overcoming these terrain-oriented challenges requires a focus on vehicle mobility, no compromises can be made in other important vehicle components such as passenger safety and protection, as well as payload and cargo requirements. With that in mind, the Army’s push toward fully autonomous vehicles is driven by the potential of autonomy to bring improvements to vehicle safety and mobility, while still meeting the necessary payload requirements.
This movement, which has inspired much of the recent research done with autonomous vehicles, was stark motivation for the detailed case study that followed, “No Driver? No Problem: Mobility Across the Autonomy Spectrum in Unmanned Ground Vehicles (UGVs),” which evaluated the mobility performance of UGVs between different levels of autonomous operation.
The study itself was multi-leveled, looking at three degrees of autonomy4 in three different types of vehicles5. In assessing the trade-offs associated with each degree of autonomy, the study concentrated on the analysis of communication-response lags in teleoperation, negotiating control priorities between humans and controllers in shared control systems, and the response of fully autonomous operation in new, unknown settings, among others. Currently, we are operating in the realm of semi-autonomous, human-robot interactive vehicles, and still have a long way before full autonomy is operational, says Dr. Paramsothy Jayakumar, Senior Research Scientist at TARDEC and case study contributor. This accentuates the importance of the control priority characteristic in shared control systems; namely, when is it alright for the robot to take over control from the human, and vice-versa? This question affects vehicle operators and passengers at all levels, from the soldier navigating the ever changing landscapes of war to the nervous sixteen-year old behind the wheel for the first time.
While many people are understandably wary of the idea of submitting their own control (especially that of a moving vehicle) to a robot, would their concerns be quelled if these robots could prevent accidents? If they could take the wheel at times of distress and create close calls out of situations that previously would’ve ended in disaster? If they could sense its human operator falling asleep and seamlessly continue driving while trying to awaken whomever is behind the wheel? If they could detect a land mine and communicate its position to surrounding vehicles, so that those vehicles would, without hesitation or second guessing, reroute to a safer path?
Before these questions could be answered, Dr. Robert Ambrose, Division Chief of the Software, Robotics and Simulation Division at NASA’s Johnson Space Center and the opening keynote speaker, took to the podium with a related and pertinent topic. His invigorating presentation of “NASA’s Human-Robotic Systems” had attendees passing around a sign-up sheet titled “Wants to Work for NASA.” The project, as Dr. Ambrose described, has roots that extend back to 1969, with limited human mobility being an issue from the outset of our very first moon landing. The Human Robotic Systems (HRS) project is working to implement advanced6 robotics in space to handle various tasks that will alleviate the human operational workload.
Dr. Hubertus Tummescheit, CEO of Modelon Inc.7, the second keynote speaker, spoke from the perspective of a small, extremely successful software company. He unveiled Modelon’s work with vehicle simulations, which respond in real time and do not compromise the model’s fidelity. This is only possible, Dr. Tummescheit described, with the use of a high level language like Modelica8 in conjunction with the Functional Mockup Interface (FMI), which is a standardized, tool independent medium used for the exchange and co-simulation of models on both system and component levels9,10.
The groundbreaking nature of this work compelled one audience member to share that this specific branch of technology was discussed as a distant dream at this same event ten years ago, and thought to not even be possible at the event fifteen years ago.
|Attendees interact with the shared-control mini Baja car that illustrates ARC research on haptic feedback for improved teleoperation. Click to enlarge.|
With the opening keynote presentations focusing on the innovative works of today, Dr. Chris Atkinson, Program Director of the Advanced Research Projects Agency-Energy (ARPA-E) and the third and final keynote speaker, urged the ARC researchers to think past even the limits that we have already set for the future. In discussing ARPA-E’s newest program area, which targets engine and powertrain technologies, Dr. Atkinson stressed that autonomy enables us to push the limits of engine fuel efficiency, even beyond the 54.5 mile per gallon fleet average fuel economy requirement set for automotive companies to be met by 2025.
In addition to the speakers mentioned, the two-day event included other presentations from TARDEC representatives, ARC Researchers (from the University of Michigan and elsewhere), and the University of Michigan Transportation Research Institute, as well as project poster viewing sessions, a teleoperated vehicle demonstration (see right), the presentation of a second case study, and laboratory tours.
The 22nd Annual ARC Program Review is scheduled for May 25-26, 2016, and will be held on The University of Michigan’s North Campus in Ann Arbor.
 Gilbert, D., and M. Hostetler. “Automotive Research Center.” 1994-95 Annual Report, Department of Mechanical Engineering and Applied Mechanics, University of Michigan 1995: 16-17. Print.
 The University of Michigan, The University of Iowa, Wayne State University, Clemson University, Oakland University, and Virginia Tech
 U.S. Army Tank Automotive Research, Development, and Engineering Center
 teleoperation, shared control, and full autonomy
 superdroid, mini Baja, HMMWV
 advanced in their individualized capabilities as well as the human-robot interaction
 Since being founded in 2005, the startup with humble beginnings has grown into an accomplished international company.
 an equation-based language used to model heterogeneous systems
 Tummescheit, H. Multi-core Real-time Simulation of High-Fidelity Vehicle Models using Open Standards: Modelica and the Functional Mockup Interface (FMI) (Abstract). Presented at the 21st Annual Automotive Research Center Program Review, May 20, 2015, Ann Arbor, MI.
 Functional Mock-up Interface. https://www.fmi-standard.org/