Resilient Fault-Tolerant Control for Multiagent UAV in Formations: Reinforcement Learning for Robust Synchronization

 

Frank L. Lewis
Moncrief-O’Donnell Endowed Chair and Head, Advanced Controls & Sensors Group
UTA Research Institute (UTARI), The University of Texas at Arlington, USA

 

 

Abstract

Small unmanned aerial vehicles (UAV) including quadrotors and multi-rotors may become the workhorses of emerging airborne delivery techniques for companies such as Amazon, FedEx, and the US Post Office.  As such, the autonomy and interactive control of such rotorcraft must reach new standards of reliability, inter-vehicle coordination, and fail-safe operation.  The benefits of multiple actuators bring with them increased risks of actuator failures and sensor noises.  Furthermore, for multiple vehicles operating in formations, the communication complexity grows exponentially and as such the allowed information flow must be strictly managed within a communication network with reduced communication interactions.

This talk will discuss some new fault-tolerant control methods for multiple rotorcraft operating in multiagent vehicle formations.  Both sensor noises and rotor actuator faults are discussed.  The quadrotor dynamics are reviewed and a two-loop design for translational dynamics and positional dynamics is given based on feedback linearization.  It is seen that if there are either sensor noises or actuator faults, additional controller dynamics are required in the form of robust observers for the leader dynamics, and estimators for the sensor and actuator states.  All of these compensator dynamics must be computed in a distributed fashion using only information from each vehicle’s neighbors in a given communication network. Design and analysis of these new control protocols is given using Lyapunov and multi-agent consensus techniques. Methods for implementing these robust fault recovery techniques are given using Reinforcement Learning, which only requires data measurements from neighbors in the graph, and no knowledge of vehicle dynamics.  Finally, a new method is given for control allocation in UAV having redundant actuators.

Biosketch

F. L. Lewis: Member, National Academy of Inventors.  Fellow IEEE, Fellow IFAC, Fellow AAAS, Fellow U.K. Institute of Measurement & Control, PE Texas, U.K. Chartered Engineer. UTA Distinguished Scholar Professor, UTA Distinguished Teaching Professor, and Moncrief-O’Donnell Chair at The University of Texas at Arlington Research Institute. Founding Member Mediterranean Control Association. Ranked at position 89 worldwide and 62 in the USA of all scientists in Computer Science and Electronics, by Guide2Research.  Bachelor’s Degree in Physics/EE and MSEE at Rice University, MS in Aeronautical Engineering at Univ. W. Florida, Ph.D. at Ga. Tech.  He works in feedback control, reinforcement learning, intelligent systems, and distributed control systems.  Author of 8 U.S. patents, 420 journal papers, 426 conference papers, 20 books, 48 chapters, and 12 journal special issues.  He received the Fulbright Research Award, NSF Research Initiation Grant, ASEE Terman Award, Int. Neural Network Soc. Gabor Award 2009, U.K. Inst. Measurement & Control Honeywell Field Engineering Medal 2009.  Received AACC Ragazzini Education Award 2018, IEEE Computational Intelligence Society Neural Networks Pioneer Award 2012 and AIAA Intelligent Systems Award 2016. IEEE Control Systems Society Distinguished Lecturer. Received Outstanding Service Award from Dallas IEEE Section, selected as Engineer of the Year by Ft. Worth IEEE Section.  Listed in Ft. Worth Business Press Top 200 Leaders in Manufacturing. Received the 2010 IEEE Region 5 Outstanding Engineering Educator Award and the 2010 UTA Graduate Dean’s Excellence in Doctoral Mentoring Award. Elected to UTA Academy of Distinguished Teachers 2012.  Texas Regents Outstanding Teaching Award 2013. He served on the NAE Committee on Space Station in 1995.