Diplomarbeit DIP-3630

Bibliograph.
Daten
Duraki, Alen: An experimental analysis of optimization base motion planning.
Universität Stuttgart, Fakultät Informatik, Elektrotechnik und Informationstechnik, Diplomarbeit Nr. 3630 (2014).
51 Seiten, englisch.
CR-Klassif.I.2.8 (Problem Solving, Control Methods, and Search)
I.2.9 (Robotics)
Kurzfassung

Through the advance of technology, robots aren’t just objects from science fiction novels, but they are becoming more and more part of our everyday life and are specially designed and constructed to support us. From little household helper that finish domestic chores to big industrial robots that assemble huge machines and automobiles, there are hundreds of different types of robots. Especially in interaction with humans it is important for a robot to react to environmental changes. In case of upcoming difficulties the robots ought not to freeze in place and take a lot of time to calculate their possibilities on how to avoid obstacles that might be in their way. Rapidly-exploring random trees and other sampling-based methods are one way to solve this problem, where robots are able to find a viable solution and execute it. However, the resulting solutions were not efficient enough and post-processing consisting of optimization methods had to be applied to smooth out the trajectories. The necessity of this optimization lead to a new generation of planning methods that were based on optimization alone. In many domains, these optimization-based motion planners constitute the state-of-the-art.

This thesis compares various optimization-based motion planners within an efficient common environmental representation with respect to the following criteria: speed, accuracy, and applicability. Through a variety of tests with arbitrary obstacles the methods will be compared and the results presented. The document concludes with an outlook on further work and possibilities.

Volltext und
andere Links
PDF (1450237 Bytes)
Abteilung(en)Universität Stuttgart, Institut für Parallele und Verteilte Systeme, Maschinelles Lernen und Robotik
BetreuerRatliff, Nathan
Eingabedatum9. September 2014
   Publ. Institut   Publ. Informatik