Engineering AI Solutions: The Future of Smarter, Safer Infrastructure

Key Concepts

Engineering AI: Purpose-built AI solutions for civil engineering and infrastructure asset management, focusing on image processing and machine learning.
Proactive and Preventive Maintenance: Shifting from a "fix it when it's broken" approach to anticipating and addressing issues before they become critical and costly.
Ditter's Law of Fives: Illustrates the exponential cost increase of delayed maintenance, emphasizing the ROI of preventative measures.
Infrastructure Asset Management: The systematic process of operating, maintaining, and upgrading infrastructure assets to maximize their lifespan and performance.
AI at Scale: Deploying AI solutions across a large number of assets, requiring careful consideration of user adoption, accessibility, and cost-effectiveness.
User-Centric Design: Tailoring technology solutions to the specific needs and workflows of end-users, rather than expecting users to adapt to the technology.
Trust and Education: Building confidence in new technologies within traditionally conservative industries through clear communication, demonstrable results, and user engagement.
Defect Identification and Severity Assessment: AI's ability to detect anomalies in infrastructure, classify them as defects, and determine their level of severity.
Defect Medical Record: A patented AI capability to track the evolution of a defect over time by associating images from different inspections, enabling the recognition of deterioration momentum.
Asset Lifespan Forecasting: Using AI to predict the remaining useful life of infrastructure assets, aiding in budget allocation and long-term planning.
Data-Driven Decision Making: Leveraging AI-generated insights to inform engineering decisions, optimize resource allocation, and justify budget requests.
Role of Engineers in the AI Era: AI is seen as a tool to augment, not replace, engineers, freeing them from tedious tasks to focus on higher-level decision-making and innovation.
Resilience and Sustainability: The increasing demand for infrastructure that can withstand environmental challenges and operate efficiently over the long term.

Dynamic Infrastructure: Deploying AI in Infrastructure Asset Management

This summary details the insights shared by Orley Ben Eliahu, COO of Dynamic Infrastructure, regarding the implementation of Artificial Intelligence (AI) across over 5,000 infrastructure assets. The discussion highlights the challenges, lessons learned, and the transformative potential of "Engineering AI" in revolutionizing how bridges, tunnels, and roads are inspected, maintained, and future-proofed.

The Need for Proactive Infrastructure Management

The conversation begins by establishing the critical need for a more proactive approach to infrastructure maintenance. The traditional "fix it when it's broken" model is unsustainable due to escalating costs and insufficient budgets, even with initiatives like the infrastructure bill. This is further underscored by Ditter's Law of Fives, which illustrates the exponential cost increase associated with delayed maintenance: a dollar spent on good construction yields $5 in preventative maintenance, $25 in repair, and $125 in full-scale rehabilitation. Dynamic Infrastructure's AI-driven solution aims to address this by enabling preventive and predictive maintenance, keeping assets in a cost-effective maintenance band.

Lessons Learned and Challenges in Deploying AI at Scale

Dynamic Infrastructure has successfully implemented AI across more than 5,000 infrastructure assets, including smaller structures like retaining walls and culverts. Orley Ben Eliahu emphasizes the most significant lesson learned: "Meet your users in their own world." This means adapting technology solutions to the needs of civil engineers, rather than expecting them to conform to the tech world. Continuous engagement with clients and users is crucial for understanding their needs and customizing the system accordingly.

The challenges encountered can be divided into two stages:

Education and Trust Building: Four years ago, AI was a buzzword, and many civil engineers were hesitant due to a lack of training and understanding of its responsible use. Dynamic Infrastructure bridged this gap through personal meetings and educational efforts.
Accessibility and User-Friendliness: The second stage involved making advanced AI technology accessible and intuitive. The goal is to provide solutions that are easy to use without extensive training or long onboarding processes, fitting within the often-restricted budgets of local governments and delivering a quick Return on Investment (ROI).

These challenges, while framed within the context of AI, are rooted in fundamental human issues of trust, budgeting, and efficiency, particularly critical when dealing with public safety and taxpayer money.

Understanding Engineering AI: Evolution and Capabilities

Engineering AI is presented as a purpose-built solution for civil engineering and infrastructure, distinct from general AI tools. Its development involves:

High-Level Image Processing and Machine Learning: The foundational technology.
Training with Civil Infrastructure Regulations and Literature: Incorporating industry-specific knowledge.
Defect Identification: Teaching the system to recognize anomalies (e.g., cracks, corrosion, vegetation growth) and classify them as defects.
Severity Assessment: Understanding the degree of damage and its potential impact.
Safety Issue Identification: Determining if a defect poses an immediate safety risk (e.g., a compromised guardrail).

This training process involves analyzing millions of images of various defects and their appearances. The system has achieved a 90.7% accuracy level in identifying defects and risks.

A key innovation is Dynamic Infrastructure's registered patent for a "medical record of a defect." This feature allows the AI to track the evolution of a defect over time by associating images from different inspections, even if taken by different people with different equipment. This enables the recognition of deterioration momentum, allowing for proactive intervention before a minor issue becomes a major, costly problem.

Furthermore, Engineering AI can forecast the condition of an asset. By analyzing a few images of a culvert, for instance, the system can estimate its remaining lifespan (e.g., needing replacement within 5 years, 5-10 years, or over 10 years). This capability empowers managers to wisely allocate budgets and justify funding requests to non-technical decision-makers by providing objective, visual evidence of asset condition and deterioration trends.

Real-World Applications and Case Studies

Flood Damage Assessment: In the aftermath of floods, Engineering AI can quickly generate before-and-after reports to visually demonstrate that bridge deterioration was caused by the event, not poor maintenance. This is crucial for securing support from organizations like FEMA. A specific example from Texas involved a county producing a report within two days after a storm to back their FEMA claim.
Budget Justification: The ability to forecast asset lifespan and demonstrate deterioration trends provides concrete data to support budget requests from legislators who may not have engineering backgrounds.
Recruitment of Young Engineers: By leveraging advanced tech tools, local governments can attract younger generations of engineers who are enthusiastic about using efficient, modern solutions, rather than being bogged down by manual paperwork.

The Evolving Role of Civil Engineers

Contrary to fears of AI replacing engineers, Orley Ben Eliahu believes that civil engineers will become happier and more efficient. Engineering AI will free them from tedious, repetitive tasks like data collection, comparison, and report generation, allowing them to focus on making smart, data-driven decisions. The responsibility for critical decisions will remain with engineers, but they will be empowered by actionable data. This shift is expected to alleviate recruitment challenges in the civil engineering sector.

The Future of Engineering AI

The future of Engineering AI is envisioned as being even more user-friendly and integrated into everyday work, akin to a "Jarvis"-like assistant. This includes:

Intuitive Interaction: Engineers will be able to communicate with their systems via voice commands, asking for specific information, risk assessments, budget summaries, and more.
Cross-Source Data Integration: AI will be capable of pulling and synthesizing relevant information from various sources.
Proactive Risk Highlighting: The system will automatically identify and highlight critical risks and issues.

Dynamic Infrastructure has already demonstrated this capability, receiving positive feedback from county engineers who were impressed by the AI's ability to understand their accents and provide concise, actionable insights. The expectation is that within five years, this level of intuitive interaction will become commonplace.

Driving Innovation in a Conservative Industry

Orley Ben Eliahu reframes the perception of the civil engineering field as "conservative" to "responsible." Given the stakes of public safety, a cautious approach is warranted. Strategies for driving innovation include:

Good Old-Fashioned Communication: Building trust through consistent dialogue and active listening.
User-Centric Development: Continuously seeking feedback from users to ensure the system serves their needs, not the other way around.
Partnership: Becoming an integral part of the users' daily work and workflows.

Advice for AEC Professionals

For civil engineers and other AEC professionals looking to lead in this technology-driven future, the advice is to:

Join the Technology: Don't just use it; help shape it.
Be Involved: Participate in conversations, share ideas and visions.
Be a Leader: Be part of the solution and the development of this new era.
Pull the Future Towards You: Proactively engage with emerging technologies rather than waiting to be impacted by them.

Conclusion

Dynamic Infrastructure's experience demonstrates that AI can be effectively deployed at scale in the infrastructure sector by prioritizing user needs, building trust through clear communication, and developing intuitive, cost-effective solutions. Engineering AI is poised to transform infrastructure asset management, making it more proactive, efficient, and data-driven, ultimately enhancing the safety and longevity of critical public assets and empowering civil engineers in their vital roles.