Open Access Repository
Exploration of the applicability of probabilistic inference for learning control in underactuated autonomous underwater vehicles



Full text not available from this repository.
Abstract
Underwater vehicles are employed in the exploration of dynamic environments where tuning of a specific controller for eachtask would be time-consuming and unreliable as the controller depends on calculated mathematical coefficients in idealisedconditions. For such a case, learning task from experience can be a useful alternative. This paper explores the capabilityof probabilistic inference learning to control autonomous underwater vehicles that can be used for different tasks withoutre-programming the controller. Probabilistic inference learning uses a Gaussian process model of the real vehicle to learn thecorrect policy with a small number of real field experiments. The use of probabilistic reinforcement learning looks for a simpleimplementation of controllers without the burden of coefficients calculation, controller tuning or system identification. A seriesof computational simulations were employed to test the applicability of model-based reinforcement learning in underwatervehicles. Three simulation scenarios were evaluated: waypoint tracking, depth control and 3D path tracking control. The 3Dpath tracking is done by coupling together a line-of-sight law with probabilistic inference for learning control. As a comparisonstudy LOS-PILCO algorithm can perform better than a robust LOS-PID. The results show that probabilistic model-basedreinforcement learning can be a deployable solution to motion control of underactuated AUVs as it can generate capablepolicies with minimum quantity of episodes.
Item Type: | Article |
---|---|
Authors/Creators: | Ariza Ramirez, W and Leong, ZQ and Nguyen, HD and Jayasinghe, SG |
Keywords: | machine learning, AUV, control, PILCO, LOS, underwater vehicle, path tracking, reinforcement learning |
Journal or Publication Title: | Autonomous Robots |
Publisher: | Springer New York LLC |
ISSN: | 0929-5593 |
DOI / ID Number: | 10.1007/s10514-020-09922-z |
Copyright Information: | © Springer Science+Business Media, LLC, part of Springer Nature 2020 |
Related URLs: | |
Item Statistics: | View statistics for this item |
Actions (login required)
![]() |
Item Control Page |