Imitation Learning based on Generative Adversarial Networks for Robot Path Planning
Robot path planning and dynamic obstacle avoidance are defined as a problem that robots plan a feasible path from a given starting point to a destination point in a nonlinear dynamic environment, and safely bypass dynamic obstacles to the destination with minimal deviation from the trajectory. Path...
Main Author: | |
---|---|
Other Authors: | |
Language: | en |
Published: |
2020
|
Subjects: | |
Online Access: | http://hdl.handle.net/10754/666096 |
id |
ndltd-kaust.edu.sa-oai-repository.kaust.edu.sa-10754-666096 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-kaust.edu.sa-oai-repository.kaust.edu.sa-10754-6660962020-11-26T05:07:58Z Imitation Learning based on Generative Adversarial Networks for Robot Path Planning Yi, Xianyong Michels, Dominik L. Computer, Electrical and Mathematical Sciences and Engineering (CEMSE) Division Wonka, Peter Moshkov, Mikhail Imitation learning mobile robot path planning generative adversarial Nets obstacle avoidance rapidly exploring random tree Robot path planning and dynamic obstacle avoidance are defined as a problem that robots plan a feasible path from a given starting point to a destination point in a nonlinear dynamic environment, and safely bypass dynamic obstacles to the destination with minimal deviation from the trajectory. Path planning is a typical sequential decision-making problem. Dynamic local observable environment requires real-time and adaptive decision-making systems. It is an innovation for the robot to learn the policy directly from demonstration trajectories to adapt to similar state spaces that may appear in the future. We aim to develop a method for directly learning navigation behavior from demonstration trajectories without defining the environment and attention models, by using the concepts of Generative Adversarial Imitation Learning (GAIL) and Sequence Generative Adversarial Network (SeqGAN). The proposed SeqGAIL model in this thesis allows the robot to reproduce the desired behavior in different situations. In which, an adversarial net is established, and the Feature Counts Errors reduction is utilized as the forcing objective for the Generator. The refinement measure is taken to solve the instability problem. In addition, we proposed to use the Rapidly-exploring Random Tree* (RRT*) with pre-trained weights to generate adequate demonstration trajectories in dynamic environment as the training data, and this idea can effectively overcome the difficulty of acquiring huge training data. 2020-11-24T12:06:00Z 2020-11-24T12:06:00Z 2020-11-24 Thesis 10.25781/KAUST-2CT9B http://hdl.handle.net/10754/666096 en 2021-11-24 At the time of archiving, the student author of this thesis opted to temporarily restrict access to it. The full text of this thesis will become available to the public after the expiration of the embargo on 2021-11-24. |
collection |
NDLTD |
language |
en |
sources |
NDLTD |
topic |
Imitation learning mobile robot path planning generative adversarial Nets obstacle avoidance rapidly exploring random tree |
spellingShingle |
Imitation learning mobile robot path planning generative adversarial Nets obstacle avoidance rapidly exploring random tree Yi, Xianyong Imitation Learning based on Generative Adversarial Networks for Robot Path Planning |
description |
Robot path planning and dynamic obstacle avoidance are defined as a problem that robots plan a feasible path from a given starting point to a destination point in a nonlinear dynamic environment, and safely bypass dynamic obstacles to the destination with minimal deviation from the trajectory. Path planning is a typical sequential decision-making problem. Dynamic local observable environment requires real-time and adaptive decision-making systems. It is an innovation for the robot to learn the policy directly from demonstration trajectories to adapt to similar state spaces that may appear in the future. We aim to develop a method for directly learning navigation behavior from demonstration trajectories without defining the environment and attention models, by using the concepts of Generative Adversarial Imitation Learning (GAIL) and Sequence Generative Adversarial Network (SeqGAN). The proposed SeqGAIL model in this thesis allows the robot to reproduce the desired behavior in different situations. In which, an adversarial net is established, and the Feature Counts Errors reduction is utilized as the forcing objective for the Generator. The refinement measure is taken to solve the instability problem. In addition, we proposed to use the Rapidly-exploring Random Tree* (RRT*) with pre-trained weights to generate adequate demonstration trajectories in dynamic environment as the training data, and this idea can effectively overcome the difficulty of acquiring huge training data. |
author2 |
Michels, Dominik L. |
author_facet |
Michels, Dominik L. Yi, Xianyong |
author |
Yi, Xianyong |
author_sort |
Yi, Xianyong |
title |
Imitation Learning based on Generative Adversarial Networks for Robot Path Planning |
title_short |
Imitation Learning based on Generative Adversarial Networks for Robot Path Planning |
title_full |
Imitation Learning based on Generative Adversarial Networks for Robot Path Planning |
title_fullStr |
Imitation Learning based on Generative Adversarial Networks for Robot Path Planning |
title_full_unstemmed |
Imitation Learning based on Generative Adversarial Networks for Robot Path Planning |
title_sort |
imitation learning based on generative adversarial networks for robot path planning |
publishDate |
2020 |
url |
http://hdl.handle.net/10754/666096 |
work_keys_str_mv |
AT yixianyong imitationlearningbasedongenerativeadversarialnetworksforrobotpathplanning |
_version_ |
1719362808590630912 |