Recent Advances in AI-Driven Automation for Geotechnical Engineering: A Review of Machine Learning Approaches

Abstract

Geotechnical engineering involves highly complex soil–structure interactions that are inherently nonlinear, heterogeneous, and uncertain in nature. Conventional analytical and empirical approaches often fail to accurately capture these complexities, as they rely heavily on simplified assumptions, laboratory testing, field investigations, and numerical modeling. These traditional methods are not only time-consuming and expensive but also depend significantly on expert judgment and interpretation, which may introduce variability in results. In recent years, artificial intelligence (AI) and machine learning (ML) techniques have emerged as transformative tools in geotechnical engineering, offering automated data analysis, pattern recognition, and improved predictive capabilities. These data-driven approaches enable engineers to model complex relationships between soil properties, environmental conditions, and structural responses more efficiently and accurately than traditional methods. This paper presents a comprehensive and systematic review of recent advancements in AI-driven automation for geotechnical engineering applications. The study follows a structured literature review methodology based on the PRISMA framework. Relevant research articles were collected from major scientific databases including Scopus, Web of Science, ScienceDirect, IEEE Xplore, and SpringerLink using targeted keywords related to machine learning applications in geotechnics. The review focuses on key application areas such as soil classification, prediction of pile and anchor capacity, tunneling-induced soil behavior, and slope stability analysis. Various machine learning models, including Artificial Neural Networks (ANN), Support Vector Machines (SVM), Random Forest (RF), Gradient Boosting algorithms (XGBoost), and deep learning architectures, are critically evaluated in terms of their predictive performance, computational efficiency, and applicability. The findings indicate that machine learning models significantly outperform traditional empirical approaches in terms of accuracy, efficiency, and adaptability. However, several challenges still hinder their widespread implementation, including limited availability of high-quality datasets, lack of standardized modeling frameworks, and issues related to model interpretability and transparency. Future research should focus on the development of hybrid physics-informed machine learning models and explainable AI techniques to enhance the reliability, robustness, and practical applicability of AI-based geotechnical systems.

Introduction

The text discusses the growing role of Artificial Intelligence (AI) and Machine Learning (ML) in geotechnical engineering, a branch of civil engineering focused on soil and rock behavior in construction projects. Traditional geotechnical methods rely on field investigations, laboratory testing, and numerical modeling, but these approaches often face challenges due to the complex, nonlinear, and variable nature of soil properties. Factors such as moisture content, mineral composition, and environmental conditions make accurate prediction difficult using conventional analytical methods.

AI and ML techniques have emerged as effective solutions because they can learn complex patterns from large datasets without requiring explicit mathematical equations. Machine learning models have been successfully applied to predict important geotechnical parameters such as soil classification, bearing capacity, settlement behavior, slope stability, and anchor pullout capacity. Advanced methods like deep learning, ensemble learning, and hybrid models have improved prediction accuracy and efficiency.

The research methodology of the study follows a systematic literature review using the PRISMA framework. Research papers were collected from major academic databases and evaluated based on machine learning techniques used, geotechnical problems addressed, dataset characteristics, model performance, and research limitations. The reviewed studies were categorized into areas such as soil behavior prediction, pile and anchor capacity prediction, slope stability analysis, and AI-based geotechnical frameworks.

The literature review highlights that deep learning models are highly effective in capturing nonlinear relationships among geotechnical parameters, while ensemble learning methods like Random Forest and Gradient Boosting often provide better accuracy and robustness than single models. AI-based approaches have also improved slope stability analysis and foundation design by reducing computational cost and increasing prediction reliability.

Overall, the studies show that AI-driven methods significantly enhance geotechnical analysis by handling complex soil behavior more effectively than traditional methods. However, challenges such as limited high-quality datasets, lack of standardized methodologies, and difficulties in interpreting AI models still limit widespread adoption. The paper concludes that AI and ML have strong potential to transform geotechnical engineering and encourages further research in this area.

Conclusion

Artificial intelligence and machine learning technologies are rapidly transforming geotechnical engineering research and practice. The reviewed studies demonstrate that AI-driven models can significantly improve prediction accuracy for soil behavior, foundation capacity, and geotechnical system performance. Machine learning approaches such as neural networks, support vector machines, ensemble models, and deep learning algorithms have been successfully applied to a wide range of geotechnical problems including pile capacity prediction, anchor pullout resistance, and slope stability analysis. Artificial intelligence and machine learning are rapidly transforming geotechnical engineering research and practice. The reviewed studies demonstrate that machine learning techniques can successfully predict soil behaviour, foundation capacity, anchor performance, and slope stability. Machine learning models offer significant advantages over traditional analytical methods, including improved prediction accuracy, reduced computational time, and the ability to handle complex nonlinear relationships. Despite these benefits, challenges such as limited datasets and model interpretability must be addressed before AI can be widely adopted in engineering practice. Future advancements in AI technologies, combined with increasing availability of geotechnical data, will likely lead to fully automated geotechnical analysis systems capable of supporting engineers in designing safer and more efficient infrastructure. Despite the promising results, further research is required to address challenges related to data availability, model interpretability, and integration with traditional geotechnical design methods. With continued advancements in AI technologies, automated and data-driven geotechnical engineering systems are expected to become increasingly common in future infrastructure projects.

References

[1] Dave, M. (2023), “Deep Learning in Geotechnical Engineering: A Comprehensive Review of Methods and Outcomes”, International Journal of Intelligent Systems and Applications in Engineering, ISSN: 2147-6799, Vol.12, pp.552-556, (DOI: 10.18201/ijisae.3826). [2] Keele, L. (2015), “Hospital Quality Risk Standardization via Approximate Balancing Weights”, Biometrics, ISSN: 0006-341X, Vol.71, pp.789-798, (DOI: 10.1111/biom.12201). [3] Aminpour, M. (2020), “Highly Efficient Reliability Analysis of Anisotropic Heterogeneous Slopes: Machine Learning Aided Monte Carlo Method”, Engineering with Computers, ISSN: 0177-0667, Vol.36, pp.1231-1246, (DOI: 10.1007/s00366-019-00794-0). [4] Aboagye, S., et al. (2021), “Energy Efficient Resource Allocation for Aggregated RF/VLC Systems”, IEEE Access, ISSN: 2169-3536, Vol.9, pp.142564-142576, (DOI: 10.1109/ACCESS.2021.3119985). [5] Talkington, S., & Claassen, M. (2023), “Linear and Non-Linear Response of Quadratic Lindbladians”, Physical Review Research, ISSN: 2643-1564, Vol.5, Article No.013051, (DOI: 10.1103/PhysRevResearch.5.013051). [6] Lin, Y., et al. (2022), “Interplay of Charge Density Wave and Magnetism on the Kagome Lattice”, Physical Review B, ISSN: 2469-9950, Vol.105, Article No.235101, (DOI: 10.1103/PhysRevB.105.235101). [7] Youwai, S., & Pamungmoon, A. (2023), “Explainable Artificial Intelligence Model for Predicting Pile Driving Vibrations”, Automation in Construction, ISSN: 0926-5805, Vol.147, Article 104733, (DOI: 10.1016/j.autcon.2023.104733). [8] Fan, L., Liu, F., & Chen, C. (2024), “Domain Adaptation of Large Language Models for Geotechnical Applications”, Computers and Geotechnics, ISSN: 0266-352X, Vol.167, Article 105742, (DOI: 10.1016/j.compgeo.2024.105742). [9] Lu, Y., Mei, G., & Piccialli, F. (2022), “Physics-Informed Extreme Learning Machine for Tunnelling-Induced Soil-Pile Interactions”, Engineering Applications of Artificial Intelligence, ISSN: 0952-1976, Vol.114, Article 104987, (DOI: 10.1016/j.engappai.2022.104987). [10] Kökçam, A. H., et al. (2025), “Predicting California Bearing Ratio with Ensemble and Neural Network Models”, Applied Sciences, ISSN: 2076-3417, Vol.15, Article 1842, (DOI: 10.3390/app15041842). [11] Bekele, Y. W. (2024), “GEOMCP: A Trustworthy Framework for AI Assisted Analytical Geotechnical Engineering”, Computers and Geotechnics, ISSN: 0266-352X, Vol.170, Article 106021, (DOI: 10.1016/j.compgeo.2024.106021). [12] Brinkgreve, R., & Zekri, A. (2021), “On the Use of Machine Learning in Geotechnical Engineering”, Computers and Geotechnics, ISSN: 0266-352X, Vol.136, Article 104207, (DOI: 10.1016/j.compgeo.2021.104207). [13] Hossain, M. S., Kim, Y., & Gaudin, C. (2018), “Experimental Investigation of Dynamically Penetrating Anchors in Clay and Silt”, Canadian Geotechnical Journal, ISSN: 0008-3674, Vol.55, pp.1242-1254, (DOI: 10.1139/cgj-2017-0415). [14] Samui, P., Kim, D., & Aiyer, B. (2012), “Pullout Capacity of Small Ground Anchor: A Least Square Support Vector Machine Approach”, Geotechnical and Geological Engineering, ISSN: 0960-3182, Vol.30, pp.1369-1378, (DOI: 10.1007/s10706-012-9546-0). [15] Shao, W., et al. (2021), “Application of Machine Learning Techniques in Geotechnical Engineering”, Mathematics, ISSN: 2227-7390, Vol.11, Article 3976, (DOI: 10.3390/math11183976). [16] Tran, V. H., et al. (2024), “Predicting Uplift Capacity of Circular Anchors Using Kolmogorov-Arnold Networks”, Engineering Applications of Artificial Intelligence, ISSN: 0952-1976, Vol.130, Article 107145, (DOI: 10.1016/j.engappai.2024.107145). [17] Li, L., et al. (2023), “Stacking Machine Learning Model for Predicting Pullout Capacity of Ground Anchors”, Computers and Geotechnics, ISSN: 0266-352X, Vol.158, Article 105346, (DOI: 10.1016/j.compgeo.2023.105346). [18] Harle, S. M., & Wankhade, R. L. (2022), “Machine Learning Techniques for Predictive Modelling in Geotechnical Engineering”, Materials Today: Proceedings, ISSN: 2214-7853, Vol.62, pp.4200-4206, (DOI: 10.1016/j.matpr.2022.04.189). [19] Ali, T., et al. (2023), “Machine Learning Architecture Replacing Laboratory Tests for Pullout Capacity of Reinforced Soils”, Computers and Geotechnics, ISSN: 0266-352X, Vol.155, Article 105129, (DOI: 10.1016/j.compgeo.2023.105129). [20] Alnmr, A., Ray, R. P., & Alsirawan, R. (2022), “State-of-the-Art Review of Reinforced Expansive Soil with Anchor Piles and Helical Piles”, Geotechnical and Geological Engineering, ISSN: 0960-3182, Vol.40, pp.5647-5665, (DOI: 10.1007/s10706-022-02098-1).

Copyright

Copyright © 2026 Prof. Harish T. Gwalani, Prof. Rushikesh R. Badnakhe, Athrava A. Nathe, Pranali V. Gulhane, Sayali S. Mohod, Yash P. Raut, Advait S. Tripathi . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.