Potholes on streets and avenues are the number one problem cited by eight out of ten Mexicans in their cities—ranking even higher than issues with drinking water supply, such as leaks and service failures—according to the results of the National Urban Public Safety Survey (ENSU) conducted by the National Institute of Statistics and Geography (INEGI).

Using advanced systems that combine computer vision cameras, laser sensors (LiDAR), and artificial intelligence models—which even leverage some smartphone features—researchers at the School of Engineering and Sciences at Tecnológico de Monterrey have developed projects to detect damage and irregularities in pavement.

The goal of these “x-rays” is to provide governments and organizations with tools to monitor road infrastructure, make accurate assessments of street conditions, and support decisions on preventive maintenance, road safety, and public policy based on precise, up-to-date evidence.

Vehicles Equipped to X-Ray the Streets

Doctoral student Luis Alejandro Arce and researcher Javier Izquierdo, based at the Mexico City campus, developed a system capable of detecting irregularities in asphalt, pinpointing their location on 3D maps, and categorizing them as potholes, speed bumps, patches, manholes, cracks, or areas with deteriorated pavement.

The project began six years ago, and Arce recalls outfitting his family’s minivan with three inertial sensors—two on the suspension and one on the chassis—using a Raspberry Pi computer and training seven AI models so the system could distinguish between different types of road irregularities.

With support from Oracle, they uploaded the data to the cloud in real time for geolocation of the events. Using a low-cost prototype (under 10,000 pesos), they managed to map 52 kilometers in the Coapa and Tlalpan District areas of Mexico City. However, they encountered one limitation: the system could only record the lane the vehicle was driving in, so they had to repeat routes on wider streets.

During a research stay at the Center for Automation and Robotics at the Polytechnic University of Madrid, Arce tested a vehicle outfitted with more advanced equipment.

Today, the equipped vehicles feature inertial sensors, cameras, computer vision algorithms, and LiDAR (Light Detection and Ranging)—a remote sensing technology that uses laser pulses to generate detailed 3D maps of streets—along with high-precision GPS.

“We were basically doing what Google Maps does: driving through the streets at low speeds—around 20 to 30 km/h—mapping everything in 3D. With just one vehicle, you can map an entire area and update it every few days. That would completely change the way repairs are planned,” says Arce.

In the next stages, the team will fine-tune the accuracy of the algorithms, reduce noise in the collected data, and retrain the models to work with different types of vehicles.

Smartphones and AI to Crowdsource Pothole Mapping

Another project, led by researcher Mahdi Zareei from the Guadalajara campus, harnesses the capabilities of smartphones—such as the accelerometer and gyroscope—along with AI models to detect potholes and record their location.

The system analyzes data from the phone’s sensors using a neural network trained to recognize vibration patterns and distinguish between potholes and speed bumps.

The goal is to implement a “mobile crowd-sensing” model, where drivers use an app to send lightweight, real-time data about pavement irregularities.

“The more users you have on the platform, the better your results,” Zareei explains. “For example, if you collect ten thousand data points, you can ignore a thousand that might contain errors, because the remaining nine thousand confirm that the pothole is there.” In its testing phase, the system reached a 98% accuracy rate.

The idea was born more than a decade ago, when Zareei was a Ph.D. student at the University of Technology in Malaysia. His goal was to create a system that could automatically detect potholes, and he later revisited the project with Carlos Alonzo López, a Master’s student in Computer Science, who carried out the system’s field tests on the streets of Mexico City.

In this experimental phase, they drove through the Cuauhtémoc and Venustiano Carranza boroughs in a sedan with an Android smartphone mounted on board, using GPS to log over 12,000 events to train and refine the system’s data.

They also designed and implemented filters to eliminate noise such as minor vibrations.

Finally, they validated the system and its data on streets in the State of Mexico, successfully identifying potholes of various sizes and distinguishing them from speed bumps.

The researcher notes that while more testing is still needed and a full platform has yet to be developed, the results so far are promising. “This is just the beginning. We could build an app where thousands of people automatically report potholes and other street issues, and that information could go directly to the government so they can respond immediately and fix them.”

Did you find this story interesting? Would you like to publish it? Contact our content editor to learn more at marianaleonm@tec.mx