Generative CoLearn: steering and cost prediction with generative adversarial nets in kinodynamic RRT

Master Thesis (2018)
Author(s)

N. Tsutsunava (TU Delft - Mechanical Engineering)

Contributor(s)

WJ Wolfslag – Mentor

Carlos Hernandez – Mentor

M. Wisse – Graduation committee member

J. Kober – Graduation committee member

Tim De Bruin – Graduation committee member

Faculty
Mechanical Engineering
Copyright
© 2018 Nick Tsutsunava
More Info
expand_more
Publication Year
2018
Language
English
Copyright
© 2018 Nick Tsutsunava
Graduation Date
05-10-2018
Awarding Institution
Delft University of Technology
Faculty
Mechanical Engineering
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Kinodynamic planning is motion planning in state space and aims to satisfy kinematic and dynamic constraints. To reduce its computational cost, a popular approach is to use sampling based methods such as RRT with off-line machine learning for estimating the steering cost and inputs. However, scalability and robustness are still open challenges in these type of Learning-RRT algorithms. We propose the use of generative adversarial networks (GAN) for learning of the steering cost and inputs. Furthermore, a novel data generation method is introduced, which is easy to learn and, in terms of parameter count, scales linearly to higher degrees of freedom. In our experiments, we show that the GAN has excellent generalisation capabilities, resulting in a considerable improvement in performance compared to the state-of-the-art. Consequently, we show that our method can scale to a planar arm and is robust to data dimensionality.

Files

Generative_colearn_appendix.pd... (pdf)
(pdf | 2.16 Mb)
- Embargo expired in 31-01-2019
License info not available
Generative_colearn_thesis_cove... (pdf)
(pdf | 2.92 Mb)
- Embargo expired in 31-01-2019
License info not available