In this talk we address two classes of challenges that can arise when implementing guidance algorithms for autonomous vehicle systems: (1) extremely large operating areas, and (2) motion that is poorly approximated by kinematic car-like models.
Our work is motivated by the desire to operate an unmanned surface vehicle (USV) in tropical riverine systems that are characterized by being poorly mapped and extremely large. As with many autonomous vehicle applications, the environment is discovered during the mission, and planning must be done in real-time. However, the environment is much too large for real-time implementation with today's fastest planning algorithms. Moreover, the motion of a USV is poorly approximated with a kinematic model since a USV displays side-slip of up to 90 degrees.
Our solution to these problems in inspired by concepts from the field of receding horizon control. In this talk, we present fundamental concepts from receding horizon control, and we discuss how they can be applied to the design of guidance algorithms for autonomous vehicles.
Dr. Dan Stilwell is an associate professor in the Bradley Department of Electrical and Computer Engineering at Virginia Tech, Blacksburg, Virginia. He earned a PhD in at Johns Hopkins University in 1999, the MS from Virginia Tech in 1993, and the BS from the University of Massachusetts in 1991. Since arriving at Virginia Tech in 2002, Dr. Stilwell has become a principal contributor to the emerging field of environmental robotics. His research spans the development of fundamental control and estimation theory for mobile sensors networks to the development of new maritime robots for both underwater and surface applications. One of his vehicle systems is now the basis of a major Navy procurement. He has been a recipient of the National Science Foundation CAREER award and the Office of Naval Research Young Investigator Program award, both in 2003. He was named a College of Engineering Outstanding Assistant Professor in 2004.