An embodied approach to the Tower of Hanoi

Keywords

Loading...
Thumbnail Image

Issue Date

2022-02-11

Language

en

Document type

Journal Title

Journal ISSN

Volume Title

Publisher

Title

ISSN

Volume

Issue

Startpage

Endpage

DOI

Abstract

Deep reinforcement techniques often simplify actions so that the actions have a direct e ect on the task. In this thesis, an embodied approach is taken instead to solve the Tower of Hanoi by having a robot learn to drive around in ve di erent environments with di erent environmental and task complexities. The robot should learn how moving actions in uence the game state. In the end, the agent is not able to learn this task even for the simplest environment, and performs even worse for higher complexities, as expected. These issues are hypothesized to be merely implementation speci c, and with adaptations to the simulation and reward signal the agent should be able to do better.

Description

Citation

Faculty

Faculteit der Sociale Wetenschappen