Publications

Detailed Information

Optimal Spin Recovery for Unmanned Aerial Vehicle Based on Reinforcement Learning : 무인기의 강화학습 기반 최적 스핀 회복

Cited 0 time in Web of Science Cited 0 time in Scopus
Authors

김동해

Advisor
김유단
Major
공과대학 기계항공공학부
Issue Date
2016-02
Publisher
서울대학교 대학원
Keywords
LoCUAVSpin recoveryBifurcation analysisReinforcement learning
Description
학위논문 (석사)-- 서울대학교 대학원 : 기계항공공학부, 2016. 2. 김유단.
Abstract
Loss-of-Control (LoC) due to an upset condition is a primary cause of aircraft accidents. LoC of unmanned aerial vehicle is harder to cope with than that of piloted aircraft because typical autopilots cannot recover the aircraft from the stable upset mode owing to the lack of control power. In this thesis, bifurcation analysis is conducted to simulate a stable flat spin mode which is the most irreparable upset. An expert system for recovering the aircraft from a flat spin mode in minimal time is also proposed. The proposed expert system consists of two phases: 1) attenuation of excessive angular velocity, and 2) stabilization to a symmetric level flight. Each phase contains an independent expert system with reinforcement learning. The performance of the expert system is compared with that of a nominal control system which imitates recovery maneuver of skilled pilots. The nominal control system is constructed by a four-step sequential recovery procedure. The optimality analysis is also performed by comparing with trajectory optimization result. Finally, the nonlinear six-degree-of-freedom simulation result is presented to demonstrate the performance of the proposed expert recovery procedure.
Language
English
URI
https://hdl.handle.net/10371/123852
Files in This Item:
Appears in Collections:

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share