Failure-Tolerant Control and Vision-Based Navigation for Hexacopters
Christopher T. Raabe · Institutional Repositories DataBase (IRDB) · 2013
The research summarized in this thesis is motivated by the events at the Fukushima Daiichi nuclear power plant that resulted from the 2011 Tohoku earthquake and tsunami.Those events exposed the need for an autonomous vehicle that can be deployed remotely to closely inspect a facility that is no longer accessible by personnel.Two key elements of this mission have been identified as requiring further research.First is the need to improve safety to avoid risk of damage or injury due to a crash while in transit to the remote location.Second is the need for a sensor that can provide position measurements with much greater precision than GPS or in environments where GPS is not available. OutlineChapter 1 introduces the hexacopter rotorcraft configuration and compares it to other types of hovering aircraft.Previous research on the subjects of fault-tolerant control of multicopters and vision-based navigation Is reviewed.Chapter 2 introduces the equations of motion that describe hexacopter flight.The various forces and moments that are applied to the hexacopter are cataloged and modeled.A linear model is developed for use in designing a simple Linear Quadratic Gaussian (LQG) controller.Chapter 3 proposes an adaptive controller that can simultaneously estimate and adapt to changes in actuator effectiveness with guarantees on speed of estimation.An error projection operator is proposed for projecting 4 error measurements to the 6 motors/propellers of the hexacopter.A method for smoothly reconfiguring upon detection iii