--- title: "Chasing zero accidents: what a year of monitoring road data taught Earthpulse and Abertis | Earthpulse" description: "Findings and insight from the Earthpulse work for the Abertis' project: Arlanda Initiative" image: https://www.earthpulse.ai/hubfs/Road%20Risk%20image.png --- [Skip to content](https://www.earthpulse.ai/resources/chasing-zero-accidents-what-a-year-of-monitoring-road-data-taught-earthpulse-and-abertis#main-content) Toggle navigation [Earthpulse](https://www.earthpulse.ai/) [search](https://www.earthpulse.ai/hs-search-results) - [Solutions](https://www.earthpulse.ai/solutions) - [Strategic](https://www.earthpulse.ai/solutions/strategic-intelligence) - [Operational](https://www.earthpulse.ai/solutions/operation-intelligence) - [Compliance](https://www.earthpulse.ai/solutions/compliance-through-satellite) - [Pulses](https://www.earthpulse.ai/solutions/pulses) - [Solutions Catalogue](https://www.earthpulse.ai/solutions-catalogue) - Products - [SPAI](https://www.earthpulse.ai/products/spai) - [Earthpulse! Labs](https://www.earthpulse.ai/earthpulse-labs) - [Sectors & Industries](https://www.earthpulse.ai/sectors-industries) - CUSTOMER STORIES - [Case Studies](https://www.earthpulse.ai/resources/tag/case-study) - Want to see how it works? - [arrow\_right\_alt Book a demo arrow\_right\_alt](https://www.earthpulse.ai/demo) - COMPANY - [About](https://www.earthpulse.ai/about-us) - [News](https://www.earthpulse.ai/resources/all) - [Contact](https://www.earthpulse.ai/contact-us) - [Insights](https://www.earthpulse.ai/resources/tag/insight) - Follow us - [Visit us on LinkedIn](https://www.linkedin.com/company/earthpulse-sl/) [Case study](https://www.earthpulse.ai/resources/tag/case-study) 26 Oct 2026 • 3 min read # Chasing zero accidents: what a year of monitoring road data taught Earthpulse and Abertis [![](https://www.earthpulse.ai/hubfs/favicon-1.ico?noresize) EarthPulse Team](https://www.earthpulse.ai/resources/author/earthpulse-team) ![Chasing zero accidents: what a year of monitoring road data taught Earthpulse and Abertis](https://www.earthpulse.ai/hubfs/Road%20Risk%20image.png?noresize) ### In this post Every year, highway networks around the world record thousands of road accidents causing injuries, fatalities, and continued operational disruption. Back in 2025, Abertis, one of the world's leading highway concession operators, launched a challenge to Earth Observation-based insight providers to find out how AI and satellite data could help get closer to its vision of *zero accidents*. [Earthpulse was selected by Abertis](https://elpais.com/proyecto-tendencias/espacio-abertis/2025-10-10/ideas-y-lugares-en-los-que-se-gesta-la-movilidad-que-viene.html) to build and validate an AI-driven road risk monitoring service, combining satellite-derived geospatial data, traffic intelligence, and weather inputs. ## The challenge Traditionally, road management companies have relied on historical accident analysis to evaluate the probability of new accidents occurring, and in the vast majority of countries around the world, road safety is mostly reactive, which means that incidents are detected once they have happened. Abertis manages over 8,600 kilometres of roads in the world. They have recognised that even on well-maintained infrastructure, risk is dynamic and it is shaped by the interaction of traffic intensity, environmental conditions, road geometry, and time of day.![arlanda-results-graphic-transparent (1)](https://www.earthpulse.ai/hs-fs/hubfs/arlanda-results-graphic-transparent%20(1).png?width=845&height=412&name=arlanda-results-graphic-transparent%20(1).png) In view of this, they want to move from reactive incident management to prevention by creating a system to identify when and where risk is building up, in near-real time, and to turn that signal into something operators can act on. ## The solution For the [Abertis challenge](https://www.abertis.com/news/earthpulse-wins-abertis-zero-accidents-ai-challenge/), Earthpulse applied the [Road Safety Pulse](https://www.earthpulse.ai/pulse/road-safety-pulse): a monitoring service that combines satellite and geospatial data, weather feeds, traffic data, and historical accident records into a model trained to classify evolving risk conditions at segment and hourly resolution. The initiative covered three highway corridors with very different geographies, data environments, and operational contexts. The highways monitored were the C-32 and AP-6 in Spain, and the Maravatío–Zapotlanejo highway (RCO) in Mexico. The model uses a particular architecture, chosen for its ability to capture non-linear interactions between road geometry, traffic, topography, and time. ![Abertis\_Screenshot6](https://www.earthpulse.ai/hs-fs/hubfs/Abertis_Screenshot6.png?width=2029&height=889&name=Abertis_Screenshot6.png) Just as important, the system was built to explain its outputs, not only to score them. An explainability layer, based on importance and feature interaction analysis, lets operators see which factors are driving a risk classification at any given moment, a key part of making the outputs trustworthy enough to act on. This is interesting not only to reveal which variables influence risk, but also how they interact with each other- for example, road curvature can have a different impact depending on whether rainfall is heavy or light. ![Abertis\_Screenshot\_2](https://www.earthpulse.ai/hs-fs/hubfs/Abertis_Screenshot_2.png?width=2032&height=888&name=Abertis_Screenshot_2.png) The solution also converts model outputs into something operators can act on directly: rather than only providing the probability outputs of an AI model, Earthpulse turns them into an [operational risk level](https://www.earthpulse.ai/solutions/operation-intelligence), defines different risk and activation thresholds together with the client and links them directly to actions, making the outputs truly actionable. ## The results By simulating how the solution would have worked in the past, across all three corridors, it showed strong alignment between predicted risk levels and observed accident patterns: around 90% of historical accidents occurred in conditions the model had classified as medium, high, or critical risk. The model is designed to flag evolving risk conditions and cannot flag or predict individual accidents. For better visualisation Earthpulse added a dedicated dashboard with a risk map covering all monitored areas, with segment-level classification.   ![Abertis\_Screenshot\_3](https://www.earthpulse.ai/hs-fs/hubfs/Abertis_Screenshot_3.png?width=2028&height=885&name=Abertis_Screenshot_3.png) The testing has produced exceptional results. In particular, around 85% of AP-6 accidents and 91% of the RCO accidents occurred in high-risk conditions or above, while C-32 achieved full accident coverage at medium risk or above. > What's compelling about this initiative isn't just the accuracy of the model developed by Earthpulse, it's what it lets us do next. The Arlanda Initiative has shown that this kind of risk monitoring works in practice, not just in theory, so we're now looking at how we bring this level of visibility across our full network and build it directly into how our teams respond day to day. The analysis through monitoring revealed that each corridor maintains a distinctive risk profile: RCO's risk was the most concentrated and varied strongly by direction, AP-6 accidents clustered in high-risk conditions calling for sustained monitoring, and C-32 behaved more as an intermediate-risk case suited to earlier, lighter intervention. That variation is itself a finding that confirms that the solution needs local calibration by corridor and direction and not a single fixed threshold. ## What's next The Arlanda Initiative has validated that the proactive road risk monitoring by Earthpulse is technically feasible. The next, pre-operational phase shifts the focus from validating the model itself to validating how it performs in daily use: integrating and deploying the model within Abertis' infrastructure, incorporating data in near-real time and embedding it fully within operators' day-to-day operations, to validate the model in live operations, assess which actions are viable and operationally feasible, and understand and measure impact over the long-term. Each corridor has its own next step. While the C-32 highway moves towards full extension and integration with Abertis' traffic management architecture. The next steps for the RCO and AP-6 will be defined at a later stage. Do you manage a road network? We can show you how Road Safety Pulse could work on your highways. [Talk to our team](https://www.earthpulse.ai/demo?utm_source=blog&utm_medium=cta&utm_campaign=abertis-case-study)   ## Share: - [Share on Facebook](https://www.facebook.com/sharer.php?u=https://www.earthpulse.ai/resources/chasing-zero-accidents-what-a-year-of-monitoring-road-data-taught-earthpulse-and-abertis) - [Share on X](https://twitter.com/intent/tweet?url=https://www.earthpulse.ai/resources/chasing-zero-accidents-what-a-year-of-monitoring-road-data-taught-earthpulse-and-abertis) - [Share on LinkedIn](https://www.linkedin.com/sharing/share-offsite/?url=https://www.earthpulse.ai/resources/chasing-zero-accidents-what-a-year-of-monitoring-road-data-taught-earthpulse-and-abertis) ## Related Articles [View All](https://www.earthpulse.ai/resources) ## [![A new evidence-based system to strengthen drought resilience in the Horn of Africa](https://www.earthpulse.ai/hubfs/The_Horn_of_Africa.jpg?noresize) News 27 May 2026 A new evidence-based system to strengthen drought resilience in the Horn of Africa Earthpulse Team](https://www.earthpulse.ai/resources/a-new-evidence-based-system-to-strengthen-drought-resilience-in-the-horn-of-africa) ## [![From space to preparedness](https://www.earthpulse.ai/hubfs/Screenshot%202026-02-20%20at%2016.44.39.png?noresize) Case study 26 Jan 2026 From space to preparedness ![false](https://www.earthpulse.ai/hubfs/favicon-1.ico?noresize) EarthPulse Team](https://www.earthpulse.ai/resources/from-space-to-preparedness) [Home Earthpulse](https://www.earthpulse.ai/) - [Visit us on LinkedIn](https://www.linkedin.com/company/earthpulse-sl/) San Juan de la Salle 42, 315 Edificio Sant Gervasi SG.2.08 08022, Barcelona [info@earthpulse.ai](mailto:info@earthpulse.ai) [For media enquiries media@earthpulse.ai](mailto:media@earthpulse.ai) [+34 628 772 669](tel:+34628772669) - Solutions - [Strategic Intelligence](https://www.earthpulse.ai/solutions/strategic-intelligence) - [Operational Intelligence](https://www.earthpulse.ai/solutions/operation-intelligence) - [Compliance](https://www.earthpulse.ai/solutions/compliance-through-satellite) - [Pulses - Custom Solutions](https://www.earthpulse.ai/solutions/pulses) - [Solutions Catalogue](https://www.earthpulse.ai/solutions-catalogue) - [See all solutions](https://www.earthpulse.ai/solutions) - Products - [SPAI](https://www.earthpulse.ai/products/spai) - [Earthpulse! Labs](https://www.earthpulse.ai/earthpulse-labs) - [Sectors & Industries](https://www.earthpulse.ai/sectors-industries) - About - [About us](https://www.earthpulse.ai/about-us) - [Contact us](https://www.earthpulse.ai/contact-us) - [Insights and stories](https://www.earthpulse.ai/resources) Copyright © EarthPulse 2026 - [Privacy Policy](https://www.earthpulse.ai/privacy-policy) - [Cookie Policy](https://www.earthpulse.ai/cookie-policy) - [Legal Notice](https://www.earthpulse.ai/legal-notice) - [Legal Text](https://www.earthpulse.ai/) - [Earthpulse! CA](https://www.earthpulse.ai/ca/homepage) Website Designed and Developed by [Syndicut](http://www.syndicut.com/) ```json { "@context" : "https://schema.org", "@type" : "BlogPosting", "author" : { "@type" : "Person", "name" : "EarthPulse Team", "url" : "https://www.earthpulse.ai/resources/author/earthpulse-team" }, "dateModified" : "2026-10-26T08:00:01.928Z", "datePublished" : "2026-10-26T08:00:01.000Z", "headline" : "Chasing zero accidents: what a year of monitoring road data taught Earthpulse and Abertis", "image" : [ "https://www.earthpulse.ai/hubfs/Road%20Risk%20image.png" ], "mainEntityOfPage" : { "@id" : "https://www.earthpulse.ai/resources/chasing-zero-accidents-what-a-year-of-monitoring-road-data-taught-earthpulse-and-abertis", "@type" : "WebPage" }, "publisher" : { "@type" : "Organization", "logo" : { "@type" : "ImageObject", "url" : "https://www.earthpulse.ai/hubfs/logo.png" }, "name" : "Earthpulse" } } ``` ```json { "@context" : "https://schema.org", "@type" : "NewsArticle", "author" : { "@type" : "Person", "name" : "EarthPulse Team" }, "dateModified" : "2026-10-26T08:00", "datePublished" : "2026-10-26T08:00", "description" : "Findings and insight from the Earthpulse work for the Abertis' project: Arlanda Initiative ", "headline" : "Chasing zero accidents: what a year of monitoring road data taught Earthpulse and Abertis", "image" : "https://147677711.fs1.hubspotusercontent-eu1.net/hubfs/147677711/Road%20Risk%20image.png", "mainEntityOfPage" : { "@id" : "https://www.earthpulse.ai/resources/chasing-zero-accidents-what-a-year-of-monitoring-road-data-taught-earthpulse-and-abertis", "@type" : "WebPage" }, "publisher" : { "@type" : "Organization", "logo" : { "@type" : "ImageObject", "url" : "//www.earthpulse.ai/hubfs/raw_assets/public/earthpulse/images/logo.png" }, "name" : "Earthpulse" } } ```