Show simple item record

dc.contributor.advisorFossen, Thor Inge
dc.contributor.authorGryte, Kristoffer
dc.date.created2015-07-01
dc.date.issued2015
dc.identifierntnudaim:12676
dc.identifier.urihttp://hdl.handle.net/11250/2352405
dc.description.abstractMotivated by the limited landing space on board a ship, this thesis investigates the landing of the Skywalker X8 fixed wing unmanned aerial vehicle (UAV) in a net. Further motivated by the way birds abruptly, and equally elegantly, reduce their velocity when landing on a perch or the branch of a tree, a perched landing strategy utilizing the increased drag experienced for large angles of attack is used to minimize the velocity at impact with the net. This is accomplished by expressing landing as an optimal control problem, taking advantage of a nonlinear model of the X8 that is valid for high angles of attack. At this stage, the optimal control problem is only concerned with the three longitudinal degreesof- freedom. It is solved in an open source optimization framework, using a nonlinear interior point method. Further, a nonlinear model predictive controller (NMPC) that can control the X8 throughout the landing is developed. At the heart of the optimal control problem lies a linear model, blended with a flat plate model to increase its validity for high angles of attack in lift, drag and pitch moment. The linear model is developed using an easy-to-use computational fluid dynamics (CFD) modeling software. In addition a six degrees-of-freedom software-in-the-loop (SITL) simulator is developed for future use in testing of hardware-near implementations of the controller, and to allow for validation of the model through pilot testing. Comparison of six different landing scenarios yield a wide range of landing velocities, depending on the constraints of the scenario. Simulations with the developed NMPC show that the same performance is achievable through control of the X8 under minor environmental disturbances. From this it is concluded that the perched landing strategy will lead to a considerable reduction in terminal absolute velocity, compared to a low angle of attack approach. It is found to be advantageous to start from a low altitude, landing into an elevated net. However, whether an equally large reduction is possible in practice depends on the capabilities of the real-time implementation and the validity of the model, particularly the propeller model, the pitching moment and drag coefficients. Finally it depends on how the lateral degrees-of-freedom are affected by the high angle of attack flight.
dc.languageeng
dc.publisherNTNU
dc.subjectKybernetikk og robotikk
dc.titleHigh Angle of Attack Landing of an Unmanned Aerial Vehicle
dc.typeMaster thesis


Files in this item

Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record