Title :
Learning hover with scarce samples
Author :
Lau, Tak Kit ; Liu, Yun-Hui
Author_Institution :
Dept. of Mech. & Autom. Eng., Chinese Univ. of Hong Kong, Hong Kong, China
Abstract :
Indoor aerial robots are useful in many applications due to their size, agility and ability to hover. However, tweaking a state-feedback controller to fly stably takes either intensive human supervision, or extensive modeling and identification, hence has never been trivial. In this paper, we give a successful flight controller design that can learn from a single demonstration performed by human and hover indoor aerial robots autonomously on maiden flight1.
Keywords :
aircraft control; autonomous aerial vehicles; control system synthesis; helicopters; hovercraft; learning (artificial intelligence); mobile robots; stability; state feedback; coaxial helicopter; extensive identification; extensive modeling; flight controller design; hover learning; indoor aerial robots; intensive human supervision; scarce samples; state-feedback controller; unmanned aerial vehicles; Cost function; Helicopters; Heuristic algorithms; Humans; Mathematical model; Robots; Stability analysis;
Conference_Titel :
Robotics and Automation (ICRA), 2012 IEEE International Conference on
Conference_Location :
Saint Paul, MN
Print_ISBN :
978-1-4673-1403-9
Electronic_ISBN :
1050-4729
DOI :
10.1109/ICRA.2012.6225165
