Current Projects

  • Discrete Event Control and Mission Planning
  • Omni-directional Vehicle Design and Control
  • Advanced Vehicle Diagnostics
  • Discrete Event Control and Mission Planning

    The tasking of a fleet of autonomous vehicles involves controlling inherently discrete actions. For example, it may be desired to achieve a set of logical tasks in a specific order. Possible tasks might include, "drop supplies at point A," "avoid obstacles," and "reach point B." The development of this type of discrete control has traditionally been achieved informally based on designer intuition and understanding. This approach has been sufficient in the past, but as complexity increases and time scales decrease, it has become apparent that formal design procedures are necessary. In the case of tasking autonomous vehicles, the problem quickly becomes complex as larger and larger numbers of vehicles are considered. Furthermore, the control strategy may need to be quickly and automatically reconfigured if environmental conditions or the mission objectives change.

    In recent years, techniques have been developed to formally describe and implement the discrete control logic needed by these classes of problems. Application of such techniques has so far been limited due primarily to the recentness of the research, but also due to problems associated with developing models as well as with computational complexity of controller verification and design. This research seeks to apply new research results to the established problem of tasking autonomous vehicles and to gain insight into how to best address these types of implementation issues. This work is supported in part by the University of Detroit Mercy Professor's Union research fund.

    Here is a sample presentation of some of the work we have done, as well as the simulation of a simple scenario.

    Other work on the formal verification and synthesis of discrete control logic is also being performed for manufacturing systems and for an automotive passice entry system (supported by TRW).

    Omni-directional Vehicle Design and Control

    The high-mobility robotic vehicle shown below has been built under a grant from the U.S. Army Tank Automotive Research, Development and Engineering Center (TARDEC). The vehicle is effectively omni-directional and draws on an active split offset castor (ASOC) design where each of three "pods" have two castored wheels. The two wheels of each pod have independent suspensions that help to maintain ground contact even over moderately rough terrain.

    The vehicle is operated remotely via a video-game type game pad that communicates wirelessly with a laptop that operates the control algorithms for the vehicle. Current work is focused on developing a control algorithm to assist the human operator in controlling the vehicle and maintaining vehicle stability and safety, even in the presence of disturbances that may arise by operating the vehicle in diverse and uneven terrains.

    Work has also been performed to develop a high-fidelity simulation of the vehicle. This simulation is being employed to safely and quickly develop control algorithms and to investigate other geometric configurations of the vehicle.

    Here is a poster from the 2012 IEEE TePRA conference showing some of the work we have done, as well as a video of some vehicle testing. Note that during testing the vehicle speed was limited to keep the vehicle safe and to avoid damaging the wheel motors which happen to be undersized for this application.

    Advanced Vehicle Diagnostics

    In collaboration with Ford Motor Company, a diagnostic design process is being developed for new and advanced vehicles where the current mainstream design process has proven to have limitations. The techniques being developed are guided by a thorough review of the application of current practices in the design of the on-board vehicle fault diagnostic systems of the first generation Ford Escape Hybrid Electric Vehicle (HEV) program and a demonstration Fuel Cell Electric Vehicle (FCEV) program.

    Based on the review and evaluation of these experiences, a preliminary tool for diagnostics design has been proposed that promises to make the design more traceable, to reduce the repetition of work, and to improve understandability and reuse. The development of this new tool is currently being undertaken at Ford.