AI-Enhanced Robotics and Autonomous Systems

Satyandra K. Gupta¹

1 Center for Advanced Manufacturing, University of Southern California, Los Angeles, CA, USA

E-mail: guptask@usc.edu

Status

The last decade has seen significant advances in AI techniques such as reinforcement learning, deep learning, large language models, and generative AI [2-6, 10]. These advances are endowing robots and autonomous systems with new capabilities. Most of the AI that we experience in our daily lives is digital AI. Examples include generating a cover letter for a job application, recommendations for watching a movie, creating a painting, and detecting a tumor in a medical image. A different kind of AI is needed to manage the behaviour of robots. For example, a robot performing sanding on an aircraft wing needs AI to operate autonomously. This AI is called physical AI. It is tasked with one or more goals, and it uses sensor data to produce a sequence of actions that the robot executes to achieve the goal. The physical AI monitors task execution using sensors and plans robot actions to perform the task. Physical AI is being used in the following areas related to robots and autonomous systems: (1) perception, (2) planning, (3) control, (4) human robot interaction, (5) learning from human demonstrations, (6) test case generation, and (7) multi robot collaborations.

Figure 1 challenges in realizing AI-powered robotic cells. It also lists advances that are necessary to

address these challenges. The risk profile of physical AI applications is often fundamentally different from that of digital AI applications. Risk consists of two aspects: (1) probability of making an error and (2) the consequence of making errors. When the consequence of making an error is not significant, then a higher probability of error can be tolerated. That is why an error probability of 1% is acceptable in many digital AI applications. Conversely, many industrial applications demand errors probabilities better than one in a million. Reducing error probability using a data-driven approach requires using enormous amounts of data. Unfortunately, acquiring data is expensive in industrial applications. Integrating model-based and data-driven approaches is needed to address the data size issue. Deployment of robotic systems takes a significant amount of human effort due to the time needed to write software and test the system. Increasing complexity of robotic systems is aggravating this problem. Unfortunately, the availability of human expertise can become a bottleneck in robot deployment. Generative AI is emerging as a tool to address this challenge. Digital twins have become a very useful tool for complex physical systems. Increasingly, AI-powered digital twins are being used to support operations of robots and autonomous systems. Finally, AI is creating new modalities for human-robot interactions.

Current and future challenges

Digital AI uses a vast amount of data during the train. Collecting high-quality data in many industrial applications takes significant time and incurs prohibitively large costs. Therefore, unfortunately, a purely data-driven AI approach is not a viable model in many industrial applications. We need physical AI to power robotics. Here are two representative use cases to show how physical AI can be used in industrial applications.

• Defect detection is an essential ingredient of robotic manufacturing. Machine learning has emerged as a powerful technique for analyzing and classifying images [11]. However, collecting a large number of images of physical defects needed to train a machine learning system is not possible. An alternative is to develop a pipeline for generating photo-realistic synthetic images. Recent work has demonstrated that a training process that utilizes a combination of photo-realistic synthetic images and real images of defects works well in practice.
• A robotic cell should be capable of building process models for new materials by autonomously conducting experiments [5,14]. While the exact quantitative relationship between the input process parameters and process performance may not be known, often qualitative relationships between many variables are known. We can utilize loss functions during the training phase that penalize deviations from known process constraints. This approach can enforce known models and accelerate the model-building process [9]. A digital twin is a digital counterpart of a real-world system [7,13]. The digital representation used in digital twins is created using data from sensors and Internet of Things devices, and it mimics the physical object or system in real-time. Digital twins are being used to provide information to task planners and

schedulers to make decisions about the next tasks to perform based on the current state of the system. Digital twins also monitor the condition and performance of machines and equipment in real-time and use this data to predict when maintenance is needed, reducing unexpected downtime and preventing machine breakdowns. To be useful in the field of robotics, digital twins need to run significantly faster than real-time. AI can be used to power the next generation of digital twins. Historically, human-robot interfaces in the industrial setting have not been very user friendly. Humans often interact with industrial robots by pressing buttons, turning knobs, and typing on keyboards. These traditional interfaces are hard to master and can be quite frustrating for a new user. Improved human-robot interfaces have potential to change the user experience and improve efficiency of the industrial operations [17,19]. Recent advances in AI are providing new ways for humans to interact with robots.

Figure 1. This figure depicts challenges in realizing AI-powered robotic cells and advances that are

necessary to address these challenges.

Advances in science and technology to meet challenges

AI is increasingly being used to augment capabilities of digital twin technology and create new capabilities to support the next generation of robotics. Here are a few examples:

• Simulations are necessary to generate optimal plans for finishing operations. Traditional simulations lack the speed required when dealing with part models with uncertainties. Machine learning is being used to create fast simulations based on neural networks, endowing digital twins with new planning and prediction capabilities.
• AI-based prognostics and health management can be used by digital twins to ensure that the onset of adverse events can be automatically detected, and corrective actions can be taken. For example, the

digital twin can utilize the force and vision data to determine the cause of rapid tool wear in robotic finishing and take corrective measures to prevent it.

Recent efforts are showing early signs of success in using generative AI in robotics applications to make humans more productive [8,15,16]. The examples below highlight opportunities for using generative AI in the field of robotics.

• Robots often need to perform complex motions to successfully execute a task. Consider the example of sanding where the robot needs to move the sanding tool in a complex motion pattern to produce a scratch-free surface finish. Generative AI now offers the capability to generate code from the text description, which enables humans to communicate with robots in a more natural, time-efficient manner and automatically create robot motion.
• Many applications require robots to perform complex tasks [12]. This requires the top-level task to be decomposed into much simpler subtasks and to determine the sequence of tasks. With the latest advancements in Large Language Models (LLMs) [20] we can pose a query such as, “Provide step-by- step directions to obtain a tool from a locked shelf.” and generate a sequence of various subtasks necessary to perform the overall task. Once atomic tasks have been identified, the robot can use a motion planner to generate the motion to execute the task. AI is revolutionizing human-machine interfaces in the following manner:
• Recent advances in natural language processing and human speech understanding are enabling new modalities for humans to interact with robots [18].
• Sometimes humans might make mistakes and ask the machine to perform an unsafe operation [1]. By monitoring human behaviors and the task state, the machine can predict occurrences of future unsafe situations and alert humans. AI can be used to simulate possible futures and perform risk assessment by accounting for uncertainties.
• Most traditional interfaces were not designed with ease of training in mind. AI-powered interfaces can provide real-time feedback, guidance, and assistance to users during the training phase, helping them navigate complex tasks or troubleshoot problems effectively. Moreover, virtual assistants equipped with AI can offer interactive support and tutorials, improving user productivity and learning outcomes during the training phase.

Concluding remarks

Physical AI needed in robotics applications cannot be realized as a monolithic system running on the cloud. Physical AI in the context of robotics should be viewed as a complex system that involves interactions among multiple AI components. The system should use the right functional decomposition to ensure that it is able to achieve the desired trade-off in performance and modularity. Many different AI approaches exist. It is unlikely that a single approach will suffice to deliver the desired performance. Therefore, each functional block should use the right AI approach by carefully considering pros and cons. Therefore, having the right system architecture in the physical AI system is the key to success in industrial applications. Generating a large amount of data is not possible in industrial applications from a time and cost perspective. Physical AI should be designed such that it can be trained with limited data generated by physical experiments. An approach that combines model-based and data-driven method is needed to successfully deploy physical AI in industrial applications. Deploying robotic cells in complex applications currently requires significant human effort. The availability of human resources needed to get this accomplished often emerges as a bottleneck and can cause delays in deployment. Generative AI is offering new tools to reduce the human expertise needed to deploy robots in industrial applications. AI-powered digital twins are ushering a new era of smart systems by lowering costs, reducing errors, improving quality, increasing performance, and reducing the environmental footprint. Humans are important parts of industrial operations and therefore

human-robot interaction issues need to be proactively addressed during the system design. AI can be used to revolutionize human-robot interfaces by promoting more intuitive interactions for workers.

Acknowledgements

This work was supported by the Center for Advanced Manufacturing at University of Southern California. I would like to thank my current and former students who contributed to this work. The authors have confirmed that any identifiable participants in this study have given their consent for publication.

References

[1] Al-Hussaini, S.; Guan, Y.; Gregory, J. M.; Pollard, K.; Khooshabeh, P.; and Gupta, S. K. (2024). Assessing the Impact of Alerts on the Human Supervisor’s Decision-Making Performance in Multi-Robot Missions. ACM Transactions on Human-Robot Interaction, 14(1): 1–40. [2] Chi, C.; Xu, Z.; Feng, S.; Cousineau, E.; Du, Y.; Burchfiel, B.; Tedrake, R.; and Song, S. (2024). Diffusion Policy: Visuomotor Policy Learning via Action Diffusion. The International Journal of Robotics Research. [3] Gao J. et al. (2024) Physically Grounded Vision-Language Models for Robotic Manipulation. 2024 IEEE International Conference on Robotics and Automation (ICRA), Yokohama, Japan, pp. 12462-12469. [4] Gregory, J. M.; and Gupta, S. K., 2024. Opportunities for Generative Artificial Intelligence to Accelerate Deployment of Human-Supervised Autonomous Robots. Proceedings of the AAAI Symposium Series, 2(1): 177–181. [5] Gupta, S.K. (2025). Embodied AI for Smart Robotic Cells in Manufacturing Applications. AAAI Conference, Senior Member Track. [6] Huang, W.; Abbeel, P.; Pathak, D.; and Mordatch, I., (2022). Language models as zero-shot planners: Extracting actionable knowledge for embodied agents. In International Conference on Machine Learning, 9118–9147. [7] Huang, Z.; Shen, Y; Li, J.; Fey, M.; and Brecher, C. (2021). A Survey on AI-Driven Digital Twins in Industry 4.0: Smart Manufacturing and Advanced Robotics. Sensors, 21(19). [8] Kang, J. H.; Dhanaraj, N.; Wadaskar, S.; and Gupta, S. K. (2024). Using Large Language Models to Generate and Apply Contingency Handling Procedures in Collaborative Assembly Applications. In 2024 IEEE International Conference on Robotics and Automation (ICRA), 15585–15592. [9] Karniadakis, G.; Kevrekidis, I. G.; Lu, L.; Perdikaris, P.; Wang, S.; Yang, L. (2021). Physics-informed machine learning. Nature Reviews Physics volume 3, pages 422–440. [10] Kusiak, A. (2020). Convolutional and generative adversarial neural networks in manufacturing. International Journal of Production Research, 58(5): 1594–1604. [11] Manyar, O. M.; Cheng, J.; Levine, R.; Krishnan, V.; Barbic, J.; and Gupta, S. K. (2022). Physics Informed Synthetic Image Generation for Deep Learning based Detection of Wrinkles and Folds. ASME Journal of Computing and Information Science in Engineering, 1–18. [12] Manyar, O. M.; McNulty, Z.; Nikolaidis, S.; and Gupta, S. K. (2023). Inverse Reinforcement Learning Framework for Transferring Task Sequencing Policies from Humans to Robots in Manufacturing Applications. In 2023 IEEE International Conference on Robotics and Automation (ICRA), 849–856. London, UK. [13] Onaji, I.; Tiwari, D.; Soulatiantork, P.; Song, B.; and Tiwari, A. (2022). Digital twin in manufacturing: conceptual framework and case studies. International Journal of Computer Integrated Manufacturing, 35(8), 831–858. [14] Patel, R.; Kanyuck, A.; McNulty, Z.; Yu, Z.; Carlson, L.; Heng, V.; Johnson, B.; and Gupta, S. K. (2024). Automated Plan Refinement for Improving Efficiency of Robotic Layup of Composite Sheets. In 2024 IEEE 20th International Conference on Automation Science and Engineering (CASE), 3132–3139. [15] Shen, W.; Garrett, Caelan; Kumar, N.; Goyal, A.; Hermans, T.; Kaelbling, L.P.; Lozano-Pérez, T.; Ramos, F. (2025). Differentiable GPU-Parallelized Task and Motion Planning. Robotics: Science and Systems, Los Angeles, California, June 21 – June 25, 2025. [16] Singh, I.; Blukis, V.; Mousavian, A.; Goyal, A.; Xu, D.; Tremblay, J.; Fox, D.; Thomason, J.; and Garg, A. (2023). ProgPrompt: program generation for situated robot task planning using large language models. Autonomous Robots, 1–14. [17] Suzuki, R.; Karim, A.; Xia, T.; Hedayati, H.; and Marquardt, N. (2022). Augmented Reality and Robotics: A Survey and Taxonomy for AR-enhanced Human-Robot Interaction and Robotic Interfaces. CHI '22: Proceedings of the 2022 CHI Conference on Human Factors in Computing Systems. [18] Tellex, S.; Gopalan, N.; Kress-Gazit, H.; and Matuszek, C. (2020). Robots That Use Language. Annual Review of Control, Robotics, and Autonomous Systems Volume 3. [19] Wiemann, R.; Posniak, L.; Pregizer, C.; and Raatz A. (2018). Intuitive Robot Programming Using Augmented Reality. Procedia CIRP, Volume 76, Pages 155-160. [20] Zhang, J.; Zhang, J.; Pertsch, K.; Liu, Z.; Ren, X.; Chang, M.; Sun, S.-H.; and Lim, J. (2023). Bootstrap Your Own Skills: Learning to Solve New Tasks with Large Language Model Guidance. In 7th Annual Conference on Robot Learning.

AI-enabled Sustainable Manufacturing

Byung Gun Joung, Albin John, and John W. Sutherland

School of Sustainability Engineering and Environmental Engineering, Purdue University, West Lafayette, USA

E-mail: bjoung@purdue.edu

Status

Artificial Intelligence (AI) is positively transforming manufacturing, and it is envisioned that one key dimension where the application represents a tremendous opportunity is AI for Sustainable Manufacturing,

i.e., AI for improved environmental performance. As global concerns over climate change, resource depletion, and environmental impact intensify, manufacturers are beginning to leverage AI technologies to optimize resource efficiency, reduce wastage, and lower carbon emissions. The application of AI to manufacturing can be a key enabler in advancing international sustainability goals such as approaching net-zero emissions and meeting the targets outlined in the UN Sustainable Development Goals (SDGs), while complementing other approaches, e.g., alternative energy adoption [1], energy efficiency improvements [2], and sustainable product design [3]. AI-enabled manufacturing is perhaps the next radical step after digital manufacturing, which seeks to computerize manufacturing. Existing manufacturing technologies, though presently limited in addressing environmental impacts and production variability, can be hyper-optimized through AI to embed environmental intelligence, enhance flexibility and scale to address the demands of a changing world. With the advancement of IoT technology and computational capabilities, Artificial Intelligence (AI) and Machine Learning (ML) are increasingly being adopted in manufacturing. In addition to their other potential performance benefits, we believe that AI/ML can accelerate the pursuit to “greener” manufacturing, e.g., decarbonization [1]. As an example, AI is being used to improve facility-wide energy efficiency by embedding real-time environmental intelligence, predictive adaptability, and scalable optimization to reduce peak demand and carbon intensity. AI plays an increasingly important role in reshaping how industries manage resources, reduce waste, and minimize environmental impact. Research on AI for sustainability not only enables data-driven analysis and learning but also calls for the development of new metrics and indicators to effectively evaluate sustainability performance [4]. Currently, AI applications in sustainable manufacturing are concentrated in a few key areas: i) Process Optimization [5]: optimizing (in real-time) resource utilization and process efficiency (e.g., highly variable demand); ii) Process Control and Quality Assurance [6]: vision systems powered by deep learning model are used to detect defects, monitor emissions, and ensure process precision—reducing rework and material waste; iii) Supply Chain Optimization [7], [8]: AI forecasts demand, manages inventories, and optimizes transportation routes, indirectly reducing emissions and resource use. Despite these successes, widespread adoption of AI in manufacturing is still limited. Many manufacturers—particularly small to medium-sized enterprises (SMEs)—face implementation barriers, e.g., lack of employee expertise, high upfront costs for infrastructure and training, concerns about data privacy, and return on investment. Moreover, many AI implementations are still focused on economic performance. The alignment of AI outcomes with environmental KPIs (e.g., carbon footprint, water use, material efficiency) is still emerging. In parallel, digital twins are becoming a crucial component to manage scalability and adaptability to handle complexity and variability in process design and optimization. These virtual models can simulate various operating conditions, material flows, equipment configurations, and uncertainties associated with real-world deployment—such as fluctuating resource availability, equipment degradation,

and process variability—enabling engineers to identify low-carbon and low-waste pathways before physical implementation. For instance, they are used to assess different production scenarios to minimize poor quality products, energy use, and chemical waste. AI can significantly improve digital twins by enhancing real-time data analysis, predictive modeling, and decision-making through advanced machine learning algorithms. AI may also be used to accelerate life cycle assessment (LCA) workflows by replacing manual inventory analysis with automated estimation based on historical data [9], product specifications [10], and production logs [11]. AI-powered LCA tools can now predict cradle-to-grave environmental impacts for new, complex designs using surrogate models trained on previously assessed products, which can easily be implemented within the design and development process to provide environmental footprint information. In materials engineering, generative models such as variational autoencoders and reinforcement learning are being applied to discover sustainable alternatives—such as bio-based polymers or recyclable alloys—that meet performance constraints while minimizing environmental burdens. These tools significantly reduce the time for R&D and cost to develop materials with less environmental impact. However, most current life cycle indicators rely heavily on predefined emission factors with various uncertainties [12], which aggregate environmental impact per unit of activity (e.g., kg CO₂ per kWh). While useful, these factors often lack spatial, temporal, and contextual granularity necessary to forecast real-world behaviors in the realm of sustainable manufacturing. As a result, they overlook site-specific environmental and health hazards associated with certain raw materials—such as toxicity, particulate emissions, heavy metal exposure, endocrine-disrupting properties, and water contamination risks—that may not be reflected in traditional GHG-focused metrics. Currently, AI in sustainable manufacturing shows great promise, but real-world implementations are isolated. Early adopters are leading the way, but a broader, systemic shift is needed to utilize the full potential of AI for sustainable manufacturing. This chapter explores current and future challenges that may hinder/delay the widespread adoption of AI in sustainability-driven manufacturing, while also identifying the gaps that must be addressed for long-term impact. It then highlights scientific and technological advances that can bridge these gaps, paving the way for transparent, adaptive, and environmentally responsible AI-enabled green manufacturing systems.

Current and future challenges

Despite their significant potential, AI and ML face structural, technological, and cultural barriers that limit their full-scale implementation in sustainable manufacturing. A primary concern for AI applications related to sustainable manufacturing is securing meaningful, relevant, and accurate data. Clean, labeled, and accessible datasets are critical for effective model training, yet many facilities operate with siloed, inconsistent, or incomplete data. Legacy systems often lack interoperability, making data integration costly and time-consuming. Additionally, concerns over intellectual property and cybersecurity create resistance to open data sharing across supply chains. The transparency, interpretability, and trustworthiness of AI models are also key issues in ensuring their effective and responsible deployment in various manufacturing applications [13]. Many state-of-the-art AI models (e.g., deep neural networks) operate as "black boxes," making it difficult for engineers and decision-makers to understand or trust their outputs completely. This limits the adoption of cross-domain and multimodal AI for tasks where accountability and traceability are crucial, such as compliance with environmental regulations or safety standards. Also, data heterogeneity and computational costs and infrastructure limitations need to be addressed to fully leverage the potential of AI in sustainable manufacturing. Another challenge with respect to adopting AI is workforce readiness, as employees need the skills to effectively use the technology [14]. The successful implementation of AI requires not only data scientists and

engineers but also skilled operators who can understand how to interpret model outputs and act upon them. Upskilling the workforce for AI-integrated environments has only occurred in a few instances, perhaps due to cost of the training/education. Recent advancement in large language models (LLMs) can support on-the- job training. These models are best utilized in general contexts but will likely struggle with more detailed/highly specialized contexts. Cross-domain and multimodal AI represents a promising frontier in advancing sustainability within manufacturing. By integrating diverse data types—such as sensor readings, textual documentation, visual inspection of images, and environmental indicators, an AI system can develop a more holistic understanding of complex manufacturing ecosystems. Also, centralized data platforms can play a critical role in coordinating domain-specific knowledge throughout the different phases of sustainability efforts. For instance, combining machine sensor data with maintenance logs and supply chain records can improve fault diagnosis, reduce material waste, and optimize energy usage across the product life cycle. Table 1 shows current application areas along with associated technologies/infrastructure needs that are essential for realizing how AI can be utilized for sustainable manufacturing.

Table 1. Manufacturing Application Areas and Associated Infrastructure /Technologies Needs

Maturity Application Area Infrastructure Needs Technology Needs

Emerging Circular Economy-Sustainability Data Hubs Cross-domain and Optimization (for scalability, Multimodal AI Green Material Discovery interoperability, and ) Agent-based Autonomous AI

Low Quality Assurance Digital Twin Real-time LCA and TEA Product Design Sensing and Actuation Explainability and Life Cycle Assessment Systems Trustworthiness of AI AI-enabled Adaptive Manufacturing

Medium

Supply Chain Optimization Predictive Maintenance

Advances in science and technology to meet challenges

Recent scientific and technological advances are crucial to overcoming challenges to AI adoption in sustainable manufacturing and such areas as energy, materials, and processes. These developments enhance efficiency, optimize resource use, and enable better monitoring and reduction of environmental impacts across the product life cycle.

One major area of progress is in AI-assisted energy optimization. Machine learning models can now analyze large volumes of sensor and operational data to dynamically control energy consumption in manufacturing systems [15]. Advanced algorithms enable real-time decision-making to reduce energy waste, schedule machinery for off-peak hours, and integrate renewable energy sources into production lines. Additionally, predictive models enhance demand forecasting and energy storage management, making industrial energy use more sustainable and resilient.

In parallel, advances in AI based digital twins and simulation technologies have revolutionized the way manufacturers design, build, operate, and evaluate systems with sustainability in mind [16]. Digital twins,

i.e., virtual representations of physical assets, allow engineers to simulate various scenarios to minimize emissions, water use, and material waste before implementation. When combined with AI, these models can adapt to changing conditions and continuously optimize performance throughout a product’s life cycle. Another key development lies in sustainable materials discovery using AI. Machine learning algorithms are accelerating the identification of low-carbon materials [17], recyclable polymers [18], and eco-friendly composites [19] by predicting material properties and behaviors from large experimental datasets. This significantly reduces the dependence on trial-and-error methods traditionally associated with material innovation and speeds up the transition to greener alternatives. Additionally, progress in AI interpretability and domain-specific modelling is bridging the gap between data science and industrial practice. New methods in explainable AI and physics-informed machine learning enable practitioners to better understand how AI models make decisions and ensure their alignment with engineering principles and sustainability goals [20]. These developments are critical for gaining trust, improving transparency, efficiency and supporting responsible adoption of AI aligned with human interaction in complex manufacturing environment. Finally, the integration of real-time AI with breakthroughs in energy systems, materials research, process simulation, and interpretability are enabling transformative improvements in green manufacturing. These scientific and technological advances are essential to overcome current challenges and ensure AI becomes a core driver of sustainable industrial development.

Concluding remarks

Artificial Intelligence (AI) has emerged as a transformative enabler in the pursuit of sustainability and green manufacturing. Its ability to analyze complex datasets, optimize resource use, and support intelligent decision-making, offers significant opportunities for reducing environmental impact across manufacturing systems. From predictive maintenance and energy-efficient scheduling to sustainable product design and supply chain transparency, AI technologies are driving operational improvements that align with long-term sustainability goals.

However, realizing the full potential of AI in this context requires more than technological readiness. It demands a multidisciplinary approach that combines data science, domain expertise, and sustainability science – in addition to, of course, manufacturing science and engineering. The successful integration of AI into manufacturing must consider not only technical performance but also explainability, data governance, and ethical implications. Additionally, it is essential to ensure that AI solutions are accessible and scalable, especially for small- and medium-sized enterprises (SMEs) that often lack the resources to adopt advanced technologies.

As industries accelerate their transition toward net-zero emissions, AI will play a growing role in enabling adaptive, transparent, and resilient manufacturing systems. The design of highly connected systems across multiple levels and layers in manufacturing can accelerate large-scale integration of AI and unleash its maximum potential. Future research should focus on advancing interpretable and centralized AI systems, integrating real-time LCA with sustainability metrics into decision-making processes, and collaborating across sectors to share knowledge and best practices. With continued innovation and responsible implementation, AI can significantly contribute to reshaping manufacturing systems into engines of sustainable development.

References

[1] Solomon BD, Krishna K. The coming sustainable energy transition: History, strategies, and outlook. Energy policy. 2011 Nov 1;39(11):7422-31, doi: https://doi.org/10.1016/j.enpol.2011.09.009. [2] Pimenov DY, Mia M, Gupta MK, Machado ÁR, Pintaude G, Unune DR, Khanna N, Khan AM, Tomaz Í, Wojciechowski S, Kuntoğlu

M. Resource saving by optimization and machining environments for sustainable manufacturing: A review and future prospects. Renewable and Sustainable Energy Reviews. 2022 Sep 1;166:112660, doi: https://doi.org/10.1016/j.rser.2022.112660. [3] Chiu MC, Chu CH. Review of sustainable product design from life cycle perspectives. International Journal of Precision Engineering and Manufacturing. 2012 Jul;13:1259-72, doi: 10.1007/s12541-012-0169-1. [4] Bachmann N, Tripathi S, Brunner M, Jodlbauer H. The contribution of data-driven technologies in achieving the sustainable development goals. Sustainability. 2022 Feb 22;14(5):2497, doi: 10.3390/su14052497. [5] Aldoseri A, Al-Khalifa K, Hamouda A. A roadmap for integrating automation with process optimization for AI-powered digital transformation. Preprints. 2023 Oct 17;1055:v1, doi: https://doi. org/10.20944/preprints202310 [6] Aragani VM. The Future of Automation: Integrating AI and Quality Assurance for Unparalleled Performance. International Journal of Innovations in Applied Sciences & Engineering. 2024;10(S1):19-27. [7] Abaku EA, Edunjobi TE, Odimarha AC. Theoretical approaches to AI in supply chain optimization: Pathways to efficiency and resilience. *International Journal of Science and Technology Research Archive.*2024 Mar;6(1):092-107. doi:

10.53771/ijstra.2024.6.1.0033 [8] Alomar MA. Performance optimization of industrial supply chain using artificial intelligence. Computational Intelligence and Neuroscience. 2022;2022(1):9306265. doi: https://doi.org/10.1155/2022/9306265 [9] Akhshik M, Bilton A, Tjong J, Singh CV, Faruk O, Sain M. Prediction of greenhouse gas emissions reductions via machine learning algorithms: Toward an artificial intelligence-based life cycle assessment for automotive lightweighting. Sustainable Materials and Technologies. 2022 Apr 1;31:e00370. doi: https://doi.org/10.1016/j.susmat.2021.e00370 [10] Kwong CK, Jiang H, Luo XG. AI-based methodology of integrating affective design, engineering, and marketing for defining design specifications of new products. Engineering Applications of Artificial Intelligence. 2016 Jan 1;47:49-60. doi: https://doi.org/10.1016/j.engappai.2015.04.001 [11] Akbar BH, Al-Aradi HJ, Achmad PR, Khan WU. Achieving Productivity and Operational Efficiency, and High-Quality Data Through Automation in Well Log Data Quality Control and Acceptance Process Using AI/ML Techniques. InAbu Dhabi International Petroleum Exhibition and Conference 2022 Oct 31 (p. D021S068R004). SPE. doi: https://doi.org/10.2118/211173-MS [12] Finnveden G. On the limitations of life cycle assessment and environmental systems analysis tools in general. The International Journal of Life Cycle Assessment. 2000 Jul;5:229-38. doi: https://doi.org/10.1007/BF02979365 [13] Soldatos J, Kyriazis D. Trusted Artificial Intelligence in Manufacturing. Boston, MA, USA: Now Publishers; 2021. doi:

10.1561/9781680838770 [14] Leesakul N, Oostveen AM, Eimontaite I, Wilson ML, Hyde R. Workplace 4.0: Exploring the implications of technology adoption in digital manufacturing on a sustainable workforce. Sustainability. 2022 Mar 11;14(6):3311. doi: https://doi.org/10.3390/su14063311 [15] Guo Y, Zhang W, Qin Q, Chen K, Wei Y. Intelligent manufacturing management system based on data mining in artificial intelligence energy-saving resources. Soft Computing. 2023 Apr;27(7):4061-76. doi: https://doi.org/10.1007/s00500-021- 06593-5 [16] He B, Bai KJ. Digital twin-based sustainable intelligent manufacturing: a review. Advances in Manufacturing. 2021 Mar;9(1):1-

21. doi: https://doi.org/10.1007/s40436-020-00302-5. [17] Mahjoubi S, Barhemat R, Meng W, Bao Y. AI-guided auto-discovery of low-carbon cost-effective ultra-high performance concrete (UHPC). Resources, Conservation and Recycling. 2023 Feb 1;189:106741. doi: https://doi.org/10.1016/j.resconrec.2022.106741 [18] Wilson AN, St John PC, Marin DH, Hoyt CB, Rognerud EG, Nimlos MR, Cywar RM, Rorrer NA, Shebek KM, Broadbelt LJ, Beckham GT. PolyID: Artificial intelligence for discovering performance-advantaged and sustainable polymers. Macromolecules. 2023 Oct 19;56(21):8547-57. doi: 10.1021/acs.macromol.3c00994 [19] Kuppusamy Y, Jayaseelan R, Pandulu G, Sathish Kumar V, Murali G, Dixit S, Vatin NI. Artificial neural network with a cross-validation technique to predict the material design of eco-friendly engineered geopolymer composites. Materials. 2022 May 10;15(10):3443. doi: https://doi.org/10.3390/ma15103443 [20] Srivastava PR, Mangla SK, Eachempati P, Tiwari AK. An explainable artificial intelligence approach to understanding drivers of economic energy consumption and sustainability. Energy Economics. 2023 Sep 1;125:106868. doi: https://doi.org/10.1016/j.eneco.2023.106868.