Give to WPI

SEMINAR - Reliable Planning and Control Algorithms for Autonomous Mobile Systems, by Dr. Raghvendra V. Cowlagi, Researcher, Aurora Flight Sciences Corporation, Cambridge, MA

Education - Colloquium - WPI Only

Thursday, February 28, 2013
11:00 AM-12:00 PM

Higgins Laboratories
HL 102

Autonomous vehicles have gained immense importance due to the broad spectrum of their military and civilian applications. Fundamental research challenges in the development of autonomous mobile vehicles include: (1) motion planning and control algorithms to enable autonomy, and (2) rigorous analyses of safety and reliability of the deployment of these vehicles within existing infrastructures, subject to human supervisory command. Novel approaches to address these challenges are discussed in this talk.

Motion planning and control of autonomous vehicles is inherently complex, because a variety of issues such as complex task specifications, dynamic environments and uncertain perception, and physical limitations of the vehicles must be addressed with limited on-board computational resources. In the first part of this talk, a new approach for optimal motion planning is discussed. The proposed motion planning approach is based on the idea of assigning transition costs to a succession of multiple edges in a graph that models the geometric characteristics of the environment. The overarching idea of geometric analyses and abstractions of vehicle motion is discussed, and other applications of such geometric analyses are discussed.

Autonomous vehicles are expected to act within existing infrastructures  such as current roads and traffic systems, and current airspace systems  under human supervisory command. The implications of introduc- ing autonomous vehicles in such infrastructures on overall system safety and reliability is ill-understood. Moreover, there is a lack of formal mathematical tools to make such safety and reliability assessments. In the second part of this talk, a new formalism for the safety and reliability analysis of large scale systems is discussed. Within this formalism, the identification of new modes of failure in large-scale hierarchical systems is shown via case studies.

In the final part of this talk, future research directions and ideas that build upon the proposed mo- tion planning framework and the proposed system safety formalism are discussed. These ideas include the tight integration of hierarchical planning and control algorithms, risk-informed motion planning, and the formulation of safety and performance metrics for future systems involving autonomous vehicles and human supervisory operators.

Suggested Audiences: College

E-mail: prheaume@wpi.edu
Phone: 508-831-5236

Last Modified: February 26, 2013 at 12:45 PM

Powered by the Social Web - Bringing people together through Events, Places, & Common Interests

 
  • Email a Friend
  • Bookmark this Page
  • Share this Page