Thor I. Fossen on "Nonlinear Observer Design and Strapdown Inertial Navigation Systems for Unmanned Aerial Vehicles"
Speaker: Professor Thor I. Fossen, Norwegian University of Science and Technology, Norway
Location: Kårhustes Hörsal
Abstract: The navigation system is one of the key components when designing modern guidance, navigation and control (GNC) systems. Navigation is usually defined as the task of determining an object’s position, velocity and attitude (PVA) based on various types of information. For decades the Kalman filter (KF), and nonlinear extensions thereof, has been used to provide integrated navigation solutions based on different types of measurements. One disadvantage of the KF is its relatively high computational complexity, with the number of internal states growing quadratically with the number of actual estimates. Another disadvantage is that the KF is developed for linear systems and stability can be difficult or impossible to prove for nonlinear extensions such as the extended Kalman filter (EKF). Consequently, there is a growing interest in the design of nonlinear observers for strapdown inertial navigation systems (INS), which can provide explicit stability guarantees and reduced computational complexity.
In this talk, a globally exponentially stable (GES) observer for attitude and gyro bias estimation is presented. The attitude observer uses gyro measurements and two or more pairs of vector measurements, typically body-fixed acceleration and magnetic field measurements. Velocity estimates from optical flow camera measurements can also be used. The attitude observer avoids the well-known topological obstructions to global stability by not confining the attitude estimate to SO(3), but rather estimating a full rotation matrix with nine degrees of freedom. A quaternion-based representation for effective implementation, which guarantees semiglobal exponential stability, is also presented. Next, a translational motion observer (TMO) for estimation of position and linear velocity is presented. The attitude and TMO observers form a feedback interconnection, which is analyzed using Lyapunov theory. Explicit stability requirements for GES are given. Low-cost integrated navigation systems using nonlinear observer theory is an emerging technology for autonomous vehicles operating in uncertain and harsh environments. The presentation focuses on the theory of nonlinear observers, stability properties and experimental validation of the methods using fixed-wing UAVs.
Biosketch: Thor I. Fossen is a naval architect and a cyberneticist. He received an MSc degree in Marine Technology in 1987 and a PhD degree in Engineering Cybernetics in 1991 both from the Norwegian University of Science and Technology (NTNU), Trondheim. He is currently professor of guidance, navigation and control and co-director of the NTNU Centre for Autonomous Marine Operations and Systems. Fossen’s expertise covers guidance systems, inertial navigation systems, nonlinear control and observer theory, vehicle dynamics, hydrodynamics, autopilots and unmanned vehicles. He has authored five textbooks. He is one of the co-founders of the DNV-GL company Marine Cybernetics where he was Vice President R&D in the period 2002-2008. He received the Automatica Prize Paper Award in 2002 and the Arch T. Colwell Merit Award in 2008 at the SAE World Congress. He has been elected to the Norwegian Academy of Technological Sciences (1998) and elevated to IEEE Fellow (2016).
This talk is a part of the ELLIIT Workshop 2017.
Spoken language: English
Cost: Free of charge
Please register at: http://www.lth.se/lu350/elliitws17/keynote2/