Minimum Risk Guidance Solutions Using Data Driven Probabilities

Share

Standard Guidance, Navigation, and Control (GN&C) systems take state data from a navigation system and create a trajectory that minimizes some a-priori determined cost function. These cost functions are typically time, money, weight, or any general physically realizable quantity. Previous work has been done to show the effectiveness of using risk as the sole objective function. However, this previous work used Poisson distributions and historical estimates to achieve this goal. In this talk we present the SRA (Situation Risk Assessment) method contained within the intelligent collision avoidance (iSC) platform. The SRA method uses data clustering, and pattern re-cognition to create a historically based estimate of guidance probabilities. These are then used in data driven, dynamic models to create the fu-ture probability fields of the situation. This probability, along with the other agent’s goals and objectives, are then used to create a minimum risk guidance solution nautical situations and for air traffic.



  Date and Time

  Location

  Hosts

  Registration



  • ETH Zürich
  • ZURICH, Switzerland
  • Switzerland 8000
  • Building: HG
  • Room Number: D 1.1
  • AIRNAV CONSULTING

  • Co-sponsored by ION-CH


  Speakers

Edwin William of The University of Southern California

Topic:

Edwin Williams, Researcher

Edwin Williams is a researcher at the University of Southern California. He has received his BS and MS degree from Purdue University, and is a PhD candidate at the University of Southern California. He has worked in the advanced program department at Pratt and Whitney, and the Guidance, Navigation, and Control group at Draper Laboratory. His work involves uncertainty quantification, machine learning, autonomous systems, and risk determination. He is a member of the AIAA, ION, USC IMPACT Laboratory, and Intel Research.

Biography:

Edwin Williams is a researcher at the University of Southern California. He has received his BS and MS degree from Purdue University, and is a PhD candidate at the University of Southern California. He has worked in the advanced program department at Pratt and Whitney, and the Guidance, Navigation, and Control group at Draper Laboratory. His work involves uncertainty quantification, machine learning, autonomous systems, and risk determination. He is a member of the AIAA, ION, USC IMPACT Laboratory, and Intel Research.

Address:The University of Southern California, , LAS ANGELES, California, United States





Agenda

Standard Guidance, Navigation, and Control (GN&C) systems take state data from a navigation system and create a trajectory that minimizes some a-priori determined cost function. These cost functions are typically time, money, weight, or any general physically realizable quantity. Previous work has been done to show the effectiveness of using risk as the sole objective function. However, this previous work used Poisson distributions and historical estimates to achieve this goal. In this talk we present the SRA (Situation Risk Assessment) method contained within the intelligent collision avoidance (iSC) platform. The SRA method uses data clustering, and pattern re-cognition to create a historically based estimate of guidance probabilities. These are then used in data driven, dynamic models to create the fu-ture probability fields of the situation. This probability, along with the other agent’s goals and objectives, are then used to create a minimum risk guidance solution nautical situations and for air traffic.