Machine learning for control of hybrid construction machines
Title: Machine learning for control of hybrid construction machines
SNIC Project: LiU-compute-2021-27
Project Type: LiU Compute
Principal Investigator: Henrique Raduenz <henrique.raduenz@liu.se>
Affiliation: Linköpings universitet
Duration: 2021-06-20 – 2022-07-01
Classification: 20399
Keywords:

Abstract

The computing resources will be used for a research project focused on development of controllers for hybrid construction machines using machine learning, more specifically reinforcement learning. Due to its trial and error nature a large number of simulations are required to obtain a controller. Also the size of the neural networks, used as function representations for the controller, are large and require a significant amount of computing when training. Therefore, the main objective with the computing resources is to speed up the process of obtaining the controllers. In this first experiment we want to get used to work with the computing resources and also improve the framework for the development of the controllers to later on apply for more resources if needed.