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.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
References
Cities: Skylines online heightmap generator. https://heightmap.skydark.pl/. Accessed 5 Apr 2023
Procedural worlds. https://www.procedural-worlds.com/. Accessed 13 Mar 2023
Barratt, S., Sharma, R.: A note on the inception score (2018)
Dey, A.: Image enhancing using ESRGAN (2019). https://github.com/tensorflow/hub/blob/master/examples/colab/image_enhancing.ipynb. Accessed 31 Mar 2023
Foundation, B.: Blender. https://www.blender.org/. Accessed 19 Mar 2023
Gasch, C., Chover, M., Remolar, I., Rebollo, C.: Procedural modelling of terrains with constraints. Multimed. Tools Appl. 79, 31125–31146 (2020)
González-Medina, D., Rodríguez-Ruiz, L., García-Varea, I.: Procedural city generation for robotic simulation. In: Robot 2015: Second Iberian Robotics Conference. AISC, vol. 418, pp. 707–719. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-27149-1_55
Goodfellow, I.J., et al.: Generative adversarial networks (2014)
Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B., Hochreiter, S.: GANs trained by a two time-scale update rule converge to a local nash equilibrium (2018)
Ho, J., Jain, A., Abbeel, P.: Denoising diffusion probabilistic models (2020)
Isola, P., Zhu, J.Y., Zhou, T., Efros, A.A.: Image-to-image translation with conditional adversarial networks. In: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 5967–5976 (2017). https://doi.org/10.1109/CVPR.2017.632
Mojang Studios: Minecraft. https://www.minecraft.net/en-us/. Accessed 11 Mar 2023
Musgrave, F.K., Kolb, C.E., Mace, R.S.: The synthesis and rendering of eroded fractal terrains. ACM Siggraph Comput. Graph. 23(3), 41–50 (1989)
OpenCV: OpenCV library. https://opencv.org/. Accessed 6 Apr 2023
Panagiotou, E., Charou, E.: Procedural 3D terrain generation using generative adversarial networks. arXiv preprint arXiv:2010.06411 (2020)
Perlin, K.: An image synthesizer. ACM Siggraph Comput. Graph. 19(3), 287–296 (1985)
Reallusion: iclone8. https://www.reallusion.com/iClone/. Accessed 19 Mar 2023
Salimans, T., Goodfellow, I., Zaremba, W., Cheung, V., Radford, A., Chen, X.: Improved techniques for training GANs (2016)
Smelik, R.M., Tutenel, T., Bidarra, R., Benes, B.: A survey on procedural modelling for virtual worlds. In: Computer Graphics Forum, vol. 33, pp. 31–50. Wiley Online Library (2014)
E-on Software: Vue - overview. https://info.e-onsoftware.com/vue/overview. Accessed 19 Mar 2023
Software, P.: Terrage. https://planetside.co.uk/. Accessed 19 Mar 2023
Tang, G.: DCGAN256. https://github.com/t0nberryking/DCGAN256. Accessed 27 Mar 2023
TensorFlow: pix2pix: image-to-image translation with a conditional GAN. https://github.com/tensorflow/docs/blob/master/site/en/tutorials/generative/pix2pix.ipynb. Accessed 3 Apr 2023
Unity Technologies: Unity real-time development platform. https://unity.com/. Accessed 27 Mar 2023
Wang, X., et al.: ESRGAN: enhanced super-resolution generative adversarial networks. In: Proceedings of the European Conference on Computer Vision (ECCV) Workshops (2018)
Acknowledgment
This work is partially supported by the Spanish Ministry of Science and Innovation under contract PID2021-124463OB-IOO, by the Generalitat de Catalunya under grant 2021-SGR-00326. Finally, the research leading to these results also has received funding from the European Union’s Horizon 2020 research and innovation programme under the HORIZON-EU VITAMIN-V (101093062) project.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Arellano, S., Otero, B., Kucner, T.P., Canal, R. (2024). A 3D Terrain Generator: Enhancing Robotics Simulations with GANs. In: Nicosia, G., Ojha, V., La Malfa, E., La Malfa, G., Pardalos, P.M., Umeton, R. (eds) Machine Learning, Optimization, and Data Science. LOD 2023. Lecture Notes in Computer Science, vol 14505. Springer, Cham. https://doi.org/10.1007/978-3-031-53969-5_17
Download citation
DOI: https://doi.org/10.1007/978-3-031-53969-5_17
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-53968-8
Online ISBN: 978-3-031-53969-5
eBook Packages: Computer ScienceComputer Science (R0)