Infrastructure as a Constant Sensor

Reflections on Structural Health Monitoring (SHM), its Future in AI, and Applications in Manufacturing

Lately, I’ve been working with Structural Health Monitoring (SHM), a technology I believe holds enormous practical potential. For anyone unfamiliar with the term, SHM focuses on continuously monitoring the structural health of buildings, bridges, and other constructions in order to detect issues or damage at an early stage.

I recently collaborated on a project where we applied SHM techniques, which was published by the Government of Mexico. You can see the full paper here (reference link).

However, beyond the technical details of that research, my main interest lies in the future potential of these methodologies—especially when combined with the rising wave of Artificial Intelligence (AI) and IoT devices.

---

A Constant Yet Unexplored Factor in the World of AI

Today, tools like Alexa, Siri, or ChatGPT rely on voice, language, and context data to provide answers. But what if they also had access to the unique “fingerprint” of every physical space? In most cases, buildings are present 99% of the time as a constant environment. If we monitor that environment properly, it could provide additional data of great value.

This doesn’t refer solely to structural information. Because the structure itself is sensitive to everything we do in our day-to-day lives—our movements, activities, and habits—we could also infer behavioral data from the patterns reflected by the structure. For example, analyzing subtle vibrations or pressure changes in certain areas of a building could yield insights into:

• The frequency and types of use for each space.

• Occupancy and mobility habits.

Imagine a future where our homes, offices, and cities offer real-time structural information that integrates with machine learning algorithms and AI. It would be as if ChatGPT not only had the ability to “listen” and “understand” what we say, but also to “feel” the space we inhabit.

---

The “Rhythm” of the Manufacturing Industry

While the vision of SHM as an environmental sensor applies to many fields, I see a huge potential for its application in the manufacturing industry, partly due to the constant rhythm of production and the need for a broad contextual understanding of operations.

Every machine, every production line, and every human-driven process maintains a cadence that, when monitored structurally, provides valuable clues to:

• Optimize efficiency.

• Anticipate potential failures.

In manufacturing, there is also a very clear commitment to understanding the details and context of operations. This facilitates adopting solutions that gather information continuously and leverage AI models to generate real-time insights, enhancing the global perspective of the production process.

---

Observations from My First Prototype

Recently, I launched a first prototype that combines SHM principles with AI techniques. During this process, I encountered an error of about 1.8% (around 2 hours 53 minutes of unmonitored time over the course of a month).

Here is how we presented the first chat that can “feel”:

The greatest value of this new approach lies in the contextual window it provides. It not only identifies the root cause of a problem but also uncovers underlying causes and patterns associated with both machine operation and staff movement. This opens up improvement opportunities that might go unnoticed with more traditional solutions.

---

The Main Promise: Automating Managerial Challenges

One of the key promises of applying these methodologies in the manufacturing industry is the automation of managerial challenges. Having a solid, pristine information base—free from human intervention and processed in real time—makes it possible to create a true “copilot” for decision-making. This copilot:

• Operates with knowledge of the entire company’s operations in real time.

• Can detect and resolve problems automatically, or at least suggest highly informed solutions, freeing personnel from repetitive tasks or continuous monitoring.

As a result, management teams can focus on strategy, innovation, and collaboration with other areas, rather than spending valuable time on tracking and resolving incidents.

---

Conclusion and Open Discussion

This integration of SHM, AI, and IoT goes beyond mere structural damage prevention. It envisions an “environmental intelligence” capable of understanding both the physical context (vibrations, deformations, pressure changes) and the operational context (human interaction, workflow). The outcome is a new level of automation, where not only machines but also infrastructure become data sources for machine learning algorithms.

• Do you think infrastructure (and its structural signals) can be a valuable data channel?

• What feasible or urgent applications do you see to harness this “physical constant” in our day-to-day lives, both in industry and other sectors?

If you’d like to explore further, feel free to check out the published paper and share your thoughts. My goal is to spark a discussion about how SHM methodologies can be integrated with new AI and IoT trends, and what the next steps are for moving from theory to practice. The idea is to create something new that allows our homes, factories, and cities to evolve technologically at the same rapid pace as software and AI itself.

I look forward to your ideas and comments!