aalto1 untyped-item.component.html
A 3D Terrain Generator: Enhancing Robotics Simulations with GANs
Loading...
Access rights
openAccess
acceptedVersion
URL
Journal Title
Journal ISSN
Volume Title
A4 Artikkeli konferenssijulkaisussa
This publication is imported from Aalto University research portal.
View publication in the Research portal (opens in new window)
View/Open full text file from the Research portal (opens in new window)
View publication in the Research portal (opens in new window)
View/Open full text file from the Research portal (opens in new window)
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Date
Major/Subject
Mcode
Degree programme
Language
en
Pages
15
Series
Machine Learning, Optimization, and Data Science - 9th International Conference, LOD 2023, pp. 212-226, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) ; Volume 14505 LNCS
Abstract
Simulation is essential in robotics to evaluate models and techniques in a controlled setting before conducting experiments on tangible agents. However, developing simulation environments can be a challenging and time-consuming task. To address this issue, a proposed solution involves building a functional pipeline that generates 3D realistic terrains using Generative Adversarial Networks (GANs). By using GANs to create terrain, the pipeline can quickly and efficiently generate detailed surfaces, saving researchers time and effort in developing simulation environments for their experiments. The proposed model utilizes a Deep Convolutional Generative Adversarial Network (DCGAN) to generate heightmaps, which are trained on a custom database consisting of real heightmaps. Furthermore, an Enhanced Super-Resolution Generative Adversarial Network (ESRGAN) is used to improve the resolution of the resulting heightmaps, enhancing their visual quality and realism. To generate a texture according to the topography of the heightmap, chroma keying is used with previously selected textures. The heightmap and texture are then rendered and integrated, resulting in a realistic 3D terrain. Together, these techniques enable the model to generate high-quality, realistic 3D terrains for use in robotic simulators, allowing for more accurate and effective evaluations of robotics models and techniques.
Description
Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
Other note
Citation
Arellano, S, Otero, B, Kucner, T P & Canal, R 2024, A 3D Terrain Generator: Enhancing Robotics Simulations with GANs. in G Nicosia, V Ojha, E La Malfa, G La Malfa, P M Pardalos & R Umeton (eds), Machine Learning, Optimization, and Data Science - 9th International Conference, LOD 2023. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14505 LNCS, Springer, pp. 212-226, International Conference on Machine Learning, Optimization, and Data Science, Grasmere, United Kingdom, 22/09/2023. https://doi.org/10.1007/978-3-031-53969-5_17