ARC Collaborative Research Seminar Series
Winter 2015

ARC seminars are free and open to the general public. Center members can download the presentation files on our password-access online portal iARC. Non-ARC members please email arc-event-inquiries@umich.edu with your requests.

If you wish to attend the seminar remotely, please contact William Lim (williamlim@umich.edu) for teleconference details. For parking information, contact Kathie Wolney (kathian@umich.edu).

Refreshments will be served 9:15-9:30am. The talks will begin at 9:30 a.m. sharp.
Venue: 1180 Duderstadt Center


January 23, Friday (9:30a.m. - 11a.m.)
University of Michigan, UM North Campus, Duderstadt Center, room 1180

Advanced Propulsion Needs
Dean McGrew, Team Leader
Wes Zanardelli, Research Electrical Engineer
Advanced Propulsion Team, U.S. Army TARDEC

        The Advanced Propulsion Team within the TARDEC Ground Vehicle Power & Mobility directorate is working to increase the level of electrification on the Army’s ground vehicle fleet to introduce new capabilities while improving overall efficiency. The team is also working to aggregate the electrical power generation and storage capability of multiple assets to facilitate microgrid and V2G capability to ensure availability of power when grid power is unstable, favorably influence power quality, optimize generator asset usage and reduce the cost of electricity. This briefing provides an overview of ongoing projects and key challenges within the team and is intended to help inspire innovative research ideas to help address these challenges.


February 13, Friday (9:30a.m. - 11a.m.)
University of Michigan, UM North Campus, Duderstadt Center, room 1180

Thrust Area 1: Dynamics and Control of Vehicles

Projects presenting:
1. Vehicle-Dynamics-Conscious Real-Time Hazard Avoidance in Autonomous Ground Vehicles (project link)
Jiechao Liu (Presenter), Jeffrey Stein (PI), Tulga Ersal (co-PI), University of Michigan; Paramsothy Jayakumar, U.S. Army TARDEC; Mitchell Rohde, Steve M. Rohde, Quantum Signal LLC

        Autonomous ground vehicles (AGVs) are gaining importance in both military and commercial fields because of their promise to navigate their environment without human intervention. Existing techniques focus either on vehicles in structured environments, or on small ground robots. In contrast, this talk focuses on obstacle avoidance for high speed, commercial-size vehicle navigation based on limited sensory range information in unknown environments. In addition to preventing the vehicle from running into obstacles, a key aspect is to ensure that avoidance maneuvers do not induce any stability or handling issues such as excessive side slip, tire lift-off, or rollover. To address this challenge, a multi-phase optimal control based nonlinear model predictive control algorithm is proposed for obstacle avoidance in high-speed, large size AGVs that perceive the environment only through a planar LIDAR sensor. For the vehicle platform of interest, the dynamical safety requirement is translated into avoiding single tire lift-off. First, the scenario where the vehicle longitudinal speed is maintained constant is investigated. Sensing and control delays are explicitly taken into consideration in the formulation to increase the robustness of the algorithm. Second, to further increase the maneuverability of the vehicle, vehicle longitudinal speed is used as the second controlled variable besides the steering angle. It is demonstrated using simulations that the proposed algorithm can navigate the AGV through the obstacle field while ensuring vehicle safety in various scenarios. Critical future directions for this work will also be discussed.

2. Improving Mobility through Latency Compensation in Teleoperated Ground Vehicles (project link)
Yingshi Zheng (Presenter), Tulga Ersal (PI), Jeffrey Stein (co-PI), Xinyi Ge, Yingshi Zheng , University of Michigan; Paramsothy Jayakumar, Mark Brudnak, U.S. Army TARDEC; Mitchell Rohde, Steve M. Rohde, Quantum Signal LLC

        Teleoperated ground vehicles are used increasingly in the Army. One important challenge with increasing the dynamic performance of teleoperated ground vehicles, such as mobility, is to handle communication delays. Previous approaches in the literature to overcome this challenge either focus on stability robustness at the expense of performance, or require system dynamics models to compensate for delays. Furthermore, these techniques are typically applied to robotic arms, and their utility in teleoperated ground vehicles is unknown.
        To compensate for the negative impact of communication latencies, a platform-independent predictor framework is being developed in this project with a specific application to teleoperated vehicles. This framework requires minimal information about the system to improve vehicle mobility and is thus beneficial to be implemented with ease on different types of vehicles with minimal tuning.
        In this talk, we will first briefly introduce the basic principles of the predictor framework. Then, we will explore in more depth the application of this framework to a simulated teleoperated vehicle within a context of velocity profile following, and demonstrate that vehicle speed error can be reduced by up to 98% with the framework.
        To further study the framework, a more comprehensive simulation environment for teleoperated vehicles is under development to test the extent the predictor framework can improve vehicle mobility under the existence of delays. In addition, a driving interface is being set up to collect human driving data. The presentation will briefly summarize these efforts, as well, and conclude with an outline of next steps.


February 27, Friday (9:30a.m. - 11a.m.)
University of Michigan, UM North Campus, Duderstadt Center, room 1180

Thrust Areas 2 Human Centered M&S and
5 Vehicle System Integration, Optimization, and Robustness

Projects presenting:
1. Teleoperation with Semi-Autonomous Behaviors and Latency (project link)
Dawn Tilbury, Justin Storms, University of Michigan

        Teleoperation of robotic assets in distant environments is plagued with difficulties including communication latency and poor perception of the robot’s environment. This project has explored how communication latency impacts human operator performance when steering a small UGV along a path. Some initial results from a user study were presented in the 2014 ARC Annual Review. This presentation will report the rest of the results from that user study. One key finding is that steering performance is dependent on the variable latency mean and standard deviation rather than the shape of the distribution. In the second part of the presentation, plans for an upcoming set of user tests on shared control will be discussed. Control will be shared between a human teleoperator and obstacle avoidance automation on board a small UGV via Model Predictive Control. Users will perform a simulated search task in an obstacle-ridden environment under different conditions of latency and prediction horizons.

2. Scaling of Cognitive Capabilities in an Integrated Cognitive-Robotic Architecture
John Laird, University of Michigan

        Current robotics systems focus on low-level perception and control to support navigation, movement, and manipulation. As we move to autonomous systems that focus on temporally extended complex task execution and coordination with humans, new computational approaches are necessary that add cognitive processing – task planning, problem solving, and human interaction. These cognitive capabilities use large bodies of real-world knowledge and experience, but must be sufficiently computationally efficient to maintain real-time reactive behavior, and they must scale to long-term task execution. In this talk, I describe our plans to study the problem of reactivity in a cognitive-robotic architecture as it scales to long-term tasks. Our approach builds on the Soar cognitive architecture (Laird 2012), which is one of the most studied and mature cognitive architectures, and one of the first integrated with real robots.


March 27, Friday (9:30a.m. - 11a.m.)
University of Michigan, UM North Campus, 2000A Phoenix Memorial Laboratory

Thrust Area 4: Advanced and Hybrid Powertrains

Jet Fuel – It’s Not Just for Planes!
Tim Edwards, Fuels Branch – Air Force Research Laboratory (brief pdf biography)

        Jet fuel is traditionally thought of as the fuel for aviation gas turbine engines. But it is commonly used in diesel engines and power generation in remote locations. Recently, interest in general aviation diesel engines has increased due to their low fuel consumption and ability to use unleaded fuel. Alternative aviation fuels from a number of processes and feedstocks have been evaluated in a variety of engines – this presentation will summarize the status of these fuels and their composition and performance. The alternative jet fuel approval process will also be reviewed.

Bulk Modulus of Compressibility Measurements of Conventional and Alternative Military Fuels (project link)
André Boehman, Taemin Kim, University of Michigan

Abstract to be posted.


ARC members can download the presentation files on our password-access online portal iARC.
Non-ARC members please email arcweb-info@umich.edu with your requests.