Martian Prints: A Voronoi-Inspired Approach to In-Situ Mars Construction within Lavatubes
Giuseppe Calabrese1 2, Jesper Van der Ploeg1* , Mo Behboodi1
1 Robotic Lab, Faculty of Architecture and the Built Environment, Delft University of Technology, Netherlands
2, International Research School of Planetary Sciences (IRSPS), University of G.d’Annunzio Pescara, Italy
Abstract
This research presents a methodology for constructing self-sustaining habitats within Martian lava tubes using regolith excavation and cementless 3D printing techniques via a swarm of minute robots. The primary objective is to establish an autarkic Design-to-Robotic-Production and Operation (D2RP&O) system for constructing extraterrestrial habitats.
To address the challenges of construction within lava tubes, this research integrates computer vision via depth sensing technology. Real-time identification and mapping of ground irregularities within the lavatube allow for optimisation of size, depth and location for the production of the habitat within the lava tube. Facilitated by parametric software, the designed structure is easily modified in response to newly revised information, ensuring optimisation of design and placements are concurrent.
The knowledge and advancements gained from this research contributes to the emerging field of space architecture, construction methodologies through cementless 3D printing and the use of autonomous robots capable of navigating the structure via ramps on optimised wall density. By establishing self-sufficient habitats in extraterrestrial environments, this research represents a contribution in expanding human presence beyond Earth.
Moreover, the application of computer vision, depth sensing technology, and parametric software extends beyond space exploration. The knowledge gained from tackling the distinct challenges posed by lava tubes can enhance construction techniques as this research not only supports future space habitation but also fosters sustainable practices and advancements in construction methodologies on Earth.
Keywords: Martian habitats, regolith excavation, cementless 3D printing, autonomous robots, sustainability.
Introduction
The exploration and colonization of Mars have captured the world's attention, as establishing self-sustaining habitats is a crucial milestone for enabling long-term human presence and future exploration missions. However, the construction of habitable structures in the harsh and inhospitable Martian environment presents formidable challenges.
In response to these challenges, the Robotic Lab of Delft University of Technology (TUDelft) has developed the research project denominated ‘Rhizome 2.0’, which introduces a methodology for constructing self-sustaining habitats within Martian lava tubes. This methodology leverages cementless 3D printing techniques inspired by Voronoi logic, combined with advanced technologies such as regolith excavation, autonomous robots, and computer vision. Through this research, promising solutions are offered to overcome the limitations of traditional construction methods on Mars.
The primary goal of our research as a team was to establish an independent Design-to-Robotic-Production and Operation (D2RP&O) system for constructing habitats on extraterrestrial bodies. This ambitious undertaking involved the development of a project that envisioned a swarm of autonomous robots, including the groundbreaking zebro rovers developed at TUDelft [1]. In this study, the exploration of alternative solutions to traditional Portland cement in concrete was taken into consideration, with a focus on the utilization of volcanic ash as a cementitious material. The utilization of volcanic ash has garnered significant interest worldwide [2][3][4][5] and was envisioned as the preferred material for 3D printing applications in our project.
The research conducted in this study not only contributes to the advancement of space exploration but also holds implications for sustainable construction practices on Earth. The insights gained from tackling the distinct challenges of constructing habitats within Martian lava tubes can enhance construction techniques and foster sustainable practices infact also in terrestrial environments. By challenging conventional construction methodologies and incorporating advanced technologies, this research represents a step towards expanding human presence beyond Earth and pushing the boundaries of innovation in construction.
In summary, the research project Rhizome 2.0, developed at the Robotic Lab of TUDelft, introduces a methodology for constructing self-sustaining habitats within Martian lava tubes. By leveraging cementless 3D printing with Voronoi logic, combined with regolith excavation, autonomous robots, and computer vision, this research provides a promising solution to overcome the limitations of traditional construction methods on Mars.
Research Methodology
The research methodology employed in this study began with the selection of a suitable Martian lava tube for habitat construction in the Tharsis Region on Mars. This selection was based on factors such as accessibility, stability, and potential for habitation. Remote sensing data and previous mission reports were carefully analyzed to identify the most suitable location, the area of Tharsis bulge was selected for its prominent population in lavatube systems and several collapsed segments that could also provide distinct pathways for access to the Martian subsurface [6].
Figure 1: Tharsis Region Lavatubes on Mars. Credit: Astrogeology science centre
Once the site was determined, the research focused on what could be the data acquisition through the deployment of autonomous robots and the respective challenges that would be faced. The robots it was assumed were specifically designed for regolith excavation and equipped with advanced sensors for the collection of data related to regolith composition, density, and physical properties. Additionally, 3d printing, depth sensing, and computer vision technologies were incorporated into a robotic arm to accurately map the irregularities within the lava tube floor and plan subsequent construction steps.
Figure 2: Identification of several challenges within the lavatube such as descent, scanning, boulder presence, irregularities of surfaces. The team concentrated on challenges highlighted in red such as scanning the surface for irregularities and respective filling to achieve a level surface for the placement of the voronoi structure.
The team started to develop a computer vision methodology that enables real-time identification and mapping of irregularities within the lava tube environment. This data informs the adaptive printing techniques employed, allowing for the creation of a level surface and optimized placement of Voronoi structures.
Figure 3: Mapping and gradual filling via 3d printing of the floor irregularities before the structures are commenced
Real-time mapping and identification of irregularities enable the adaptive printing approach, adjusting to the specific conditions of each irregularity to be filled in order to achieve a level surface; this not only enhances the structural integrity of the habitat but also optimizes material distribution and subsequent construction processes. Not being able to source hundreds of photos of lavatube ground surfaces, a popular procedural generation algorithm in parametric software denominated ‘perlin noise’ was utilized to generate textures and terrain synthetically to train a computer vision model.
Figure 4: Computer vision synthetic data generation via the algorithm perlin noise to image generation
Assuming a vertical camera was positioned on the robotic arm of a rover, directly vertically to the ground irregularities, several hundreds of coloured images were extracted, from there 80% where utilized for training the model and 20% for testing purposes in order to extract depth maps from future images as a model output prediction.
Figure 5: Computer Vision model training and model output prediction
The collected data served as the foundation for training a computer vision model in the research. Python code was utilized to implement a methodology that simplified the 512x512 pixel images obtained from the Grasshopper Rhino software. This involved applying a thresholding technique to generate masks representing five distinct levels of pouring regolith, with darker regions indicating earlier steps and progressively lighter regions representing deeper levels. These masks, along with colored ground truth images, constituted the training dataset. The model was trained to accurately segment and classify the different levels of pouring regolith based on the masks. It utilized the masks to guide the 3D printing process by identifying darker areas, corresponding to deeper regions on the ground surface irregularities, ensuring the creation of a level and structurally stable surface for subsequent construction activities within the lavatube environment.
The trained PyTorch model for mask prediction demonstrates a training accuracy of 0.8, while the evaluating accuracy stands at 0.4. Although the training accuracy appears promising, the low evaluating accuracy suggests a lack of performance on unseen data. To enhance the model, several steps can be taken. These include augmenting the training data to increase diversity and quantity, acquiring more labeled data for improved generalization, re-evaluating the model architecture, tuning hyperparameters, applying regularization techniques, experimenting with optimization algorithms, employing ensemble learning, performing error analysis, and regularly evaluating the model's performance. Continuous iteration and refinement of these steps will aim to achieve higher evaluating accuracy and enhance the model's ability to predict masks effectively.
Figure 6: Training Accuracy and Validation Accuracy
Completed the computer vision surface regularity challenge, the focus was the design of the voronoi logic structure. Voronoi structures offer numerous advantages in architecture and design. They optimize material usage, resulting in lightweight and efficient structures. The unique and visually appealing patterns enhance aesthetics while allowing for customization and adaptability. Inspired by nature, Voronoi structures promote sustainability by reducing material waste. They also provide structural resilience and cost-effectiveness. Overall, Voronoi structures combine functionality, aesthetics, and sustainability in innovative architectural solutions [7]. A diagram of functions and spaces was developed and translated into voronoi logic via parametric software.
Figure 7: Functional diagram space layout for four people and voronoi development scheme
A pipe line was developed towards achieving the voronoi logic habitat design commencing from parametric software to the structural analysis.
Figure 8: The work pipeline started with the creation of space relationship amongst functions that was then translated into a structure with a voronoi logic and analysed structurally.
Figure 9: An interesting wall junction fragment was isolated.
Figure 10: The wall fragment was optimised with variable stiffness of voronois. The optimization was tailored towards the introduction of a ramp as a path for the rover for 3d printing. This was achieved via printing denser areas of voronois in the wall. The structural optimisation was not integrated into the optimization due to lack of time.
A printing path was developed within the structurally optimised wall, via the incorporation of variations in density of the Voronoi patterns. The voronoi allow for the possibility of integrating piping, furniture and ECLSS [8] also in the future.
The printing process of the rover is via an innovative upward spiral movement, taking inspiration from the spiral ramp construction methodology theorized by French architect Jean-Pierre Houdin for the pyramids [9].
Figure 11 : Spiral ramp development stages along the voronoi elements, integrated within the wall structure. The blue line along the wall illustrates the rover passage for production.
Figure 12: The dark line along the walls illustrates the rover passage for production over the ramps with its mechanical arm. The rover would print at times in forward direction and at times in reverse but never double printing on the same trajectory to avoid creation of internal steps.
This approach, guided by the principles of Design-to-Robotic-Production and Operation (D2RP&O) specifically developed for extraterrestrial construction, ensures efficient and precise construction of the habitat. Furthermore, an achievement of this research is the integration of a staircase within the wall structure, demonstrating significant advancements in design and construction techniques. This integration enhances accessibility and functionality within the habitat, showcasing the ingenuity and practicality of the developed methodology.
Figure 13: Structural optimisation of a selected wall fragment with densification of voronoi, and overlay of the spiral ramp path for the 3d printing rover throughout the structure.
Figure 14: Robotic arm and rover radius restrictions in 3d printing
Beyond its immediate implications for space exploration, this research also has significant potential to enhance construction practices and promote sustainability on Earth. The knowledge and innovations derived from this study can be applied to optimize construction processes, improve structural design, and drive sustainability on Earth.
By revolutionizing construction methodologies both within and beyond Earth, this research represents a crucial step forward in expanding human presence and establishing self-sufficient habitats in extraterrestrial environments. The integration of cutting-edge technologies and the development of adaptive printing techniques using computer vision and depth sensing set the stage for the realization of our interplanetary ambitions.
Conclusions
In this study, we have presented a methodology for constructing self-sustaining habitats within Martian lava tubes utilizing cementless 3D printing.
The major outcome of this research is the successful integration of computer vision technology to address the challenge of creating a level surface within lava tubes and the scaling up via the inclusion of a printing method achieved via the spiral ramp concept. Through training and implementation, our computer vision system was able to accurately determine the depth and effectively close voids in the floor surface. This achievement is a significant advancement in adapting construction techniques to irregular environments and lays the foundation for precise and stable placement of Voronoi structures. Based on the findings of this study, several recommendations can be made for future research and practical implementation. Firstly, continuous refinement and testing of the cementless blend used in 3D printing are essential to ensure its suitability for long-term habitation on Mars. Additionally, further advancements in autonomous robotics technology should be pursued to optimize the excavation process and improve the scalability and efficiency of construction operations.
Moreover, expanding the application of computer vision technology to other aspects of habitat construction, such as structural inspection and maintenance, can enhance the overall safety and reliability of extraterrestrial habitats. Lastly, collaboration and knowledge-sharing among researchers, space agencies, and industry stakeholders are vital to accelerate progress in the field of space architecture and enable the realization of self-sustaining habitats beyond Earth.
Study Limitations
While this research has achieved milestones, it is important to acknowledge certain limitations. Firstly, the efficacy of the cementless blend used in 3D printing Martian regolith requires investigation and optimization to ensure long-term durability and structural integrity in the harsh Martian environment. Additionally, we believe the scalability and efficiency of the autonomous robot swarm in large-scale excavation projects needs to be explored and enhanced particularly regarding the design of the wheels, the robotic arm and the interaction with the pumping/mixer following and assisting the rovers along their printing paths. Furthermore, due to certain constraints, we were unable to physically print and test our model, limiting our ability to identify and address potential issues that may arise in the construction process and harsh environment such as delamination and cracking between layers due to thermal stress or not optimized blend.
Moreover, the application of our methodology is primarily focused on Martian lava tubes, and its transferability to other extraterrestrial environments will require additional adaptations and considerations. It is crucial to conduct further research and testing to evaluate the feasibility and limitations of this approach in diverse extraterrestrial scenarios, such as different planetary surfaces, geological compositions, and environmental conditions. By expanding the scope of our investigations, a more comprehensive understanding of the challenges and constraints involved can be achieved, facilitating the development of tailored solutions for each unique extraterrestrial environment.
Acknowledgements
Henriette Bier
Arwin Hidding
Casper van Engelenburg
Seyran Khademi
J.M. Prendergast
References
1. Marc et al, “First results and Lessons Learned fo CHILL-ICE 2021 Field Campaign”, EGU22, the 24th EGU General Assembly, held 23-27 May, 2022 in Vienna, Austria and Online. May 2022. DOI: 10.5194/egusphere-egu22-611
2. Coppola et al, “Binder Alternative to Portland Cement and waste management for sustainable construction”, Journal of Applied Biomaterials & Functional Materials, Vol 16(3), pp. 186-202, Jul 2018. https://doi.org/10.1177/228080001878
3. Djobo et al, “Volcanic ash-based geopolymer cements/concretes: the current state of the art and perspectives”, Journal of Environmental Science and Pollution Research, Vol 24, pp. 4433–4446, Dec 2016. https://doi.org/10.1007/s11356-016-8230-8
4. Nazneen et al, “Geopolymer: A potential alternative binder for the sustainable development of concrete without ordinary Portland cement, Journal of Industrial Pollution Control”, Vol.33, pages 1500-1504, Nov 2017. https://www.semanticscholar.org/paper/GEOPOLYMER-A-POTENTIAL-ALTERNATIVE-BINDER-FOR-THE-Nazneen-Daggubati/b1095cd5b0f0354e254a7498138327d64960a4a5
5. Schiavone et al, “Pozzolan Based 3D Printing Composites From the Formulation”, Vol 14(1), page 43, Journal of Materials, Dec 2020. https://doi.org/10.3390/ma14010043
6. Crown et al, “Distribution and Morphology of Lava Tube Systems on the Western Flank of Alba Mons, Mars. JGR Planets. June 2022. https://doi.org/10.1029/2022JE007263
7. Sahin et al, “Examining the use of voronoi diagrams in architecture on a student project”, 3rd International Conference on New Trends in Architecture and Interior Design, Proceedings book, April 2017.
8. Hiding et al, “Structural adaptation through stiffness tuning”. Spool, 6(1 #2 ), 43-48, 2019. https://doi.org/10.7480/spool.2019.1.3892
9. Houdin et al, “La construction de la pyramide de Khéops : vers la fin des mystères?”. Annales des Ponts et Chaussees, Elsevier. Jan-Mar 2022. https://doi.org/10.1016/S0152-9668(02)80012-2