ROADVIEW Advances Automated Driving in Rain, Fog and Snow

European Research Project Demonstrates Robust Perception, Localisation and Connected Intelligence for Automated Vehicles in Adverse Weather Conditions

Rain, fog, snow and low visibility remain among the greatest challenges for automated vehicles. Adverse weather conditions can affect sensor performance, reduce perception range and compromise the reliability of driving decisions. The Horizon Europe innovation project ROADVIEW [1](Robust Automated Driving in Extreme Weather) has addressed these challenges and presented its results after four years of research and development at its final event hosted at the AVL Mobility and Sensor Center Roding in Germany, on August 19, 2026.

With a budget of €9.7 million, ROADVIEW brought together 15 partners from industry and academia across seven European countries. The project’s objective was to develop robust and cost-efficient perception and decision-making systems for automated vehicles capable of operating reliably under challenging environmental conditions.

At the heart of the project was a key question: How can an automated vehicle not only understand its environment, but also assess how reliable its sensor information is under changing weather conditions?

To address this challenge, ROADVIEW developed an integrated system architecture that combines perception, localisation, decision-making and validation. The technologies developed within the project include Adaptive Sensor Fusion, 3D Outlier Detection, Road Grip Estimation, Visibility Detection, HD Mapping and EA-NDT Localisation, weather-conditional navigation and velocity control, as well as infrastructure-based perception and V2X communication. These developments were complemented by simulation-assisted testing and validation methods designed to accelerate the verification of automated driving functions in adverse weather conditions.

ROADVIEW developed a broad portfolio of technologies covering the entire perception, localisation and decision-making chain. Three examples illustrate the project’s approach particularly well, as demonstrated in the final ROADVIEW event in Germany.

One example is 3D Outlier Detection for LiDAR data. Rain, fog and snow can generate additional sensor returns that do not originate from real objects and can negatively impact perception performance. ROADVIEW developed methods to identify and remove these weather-induced artifacts from LiDAR point clouds. Project investigations showed that the filtering approach can improve downstream tasks, including semantic segmentation by approximately 5% and object detection by up to 2%.

Another focus area was robust localisation. Automated vehicles require reliable positioning even when satellite-based navigation signals are unavailable. ROADVIEW developed a LiDAR-based localisation approach that matches live sensor data against a high-definition, semantically enriched representation of the environment, enabling reliable positioning in challenging conditions.

Beyond vehicle-based perception, ROADVIEW also investigated how infrastructure and vehicles can work together more effectively. Through V2X communication, infrastructure sensors can provide additional information about objects, weather conditions and the surrounding environment. This creates an extended environmental model that enhances vehicle situational awareness, particularly under reduced visibility conditions.

The consortium made a significant contribution to the development and validation of ROADVIEW technologies. Activities included collecting extensive sensor data under real-world weather conditions throughout all four seasons,

temperature conditions ranging from –10°C to 50°C.

At the AVL Mobility and Sensor Center Roding, the ROADVIEW technologies were evaluated under controlled and repeatable conditions. The facility’s weather hall can reproduce rainfall intensities of up to 120 millimeters per hour, dense fog with visibility reduced to approximately seven meters, and a wide range of lighting conditions, including nighttime scenarios. This enabled the consortium to systematically assess system performance under critical environmental conditions.

We refer readers to the official ROADVIEW YouTube channel[1], which features a variety of demonstration videos showcasing the project’s key technologies, validation activities, and automated driving performance under adverse weather conditions.

In addition to its technical innovations, the ROADVIEW consortium has released several datasets collected under harsh weather conditions. Among these, Snowy Scenes[2] is a unique multimodal dataset recorded entirely in real-world snowy environments in Finland. The dataset contains more than 22,000 synchronised multimodal samples and supports multiple perception tasks, including 3D object detection, semantic segmentation, and point cloud denoising/filtering, making it a valuable resource for advancing robust autonomous driving in winter conditions.

“ROADVIEW aims to go from autonomous to snow-tonomous by developing a unique approach to automated mobility, allowing more powerful and reliable onboard perception and control technologies capable of working under severe environmental conditions such as snow, rain, and fog. As the ROADVIEW project coordinator, I am proud that we have successfully advanced automated driving from operating mainly in fair weather toward reliable performance in such challenging weather conditions. Through the development of robust multimodal perception, weather-aware localisation, intelligent decision-making, V2X-supported operation, and innovative X-in-the-Loop validation methodologies, ROADVIEW has demonstrated how connected and automated vehicles can achieve higher safety, reliability, and resilience under adverse weather conditions. Our results provide important scientific, technological, and regulatory foundations for the future deployment of trustworthy automated mobility systems across Europe.” says Assoc. Prof. Eren Aksoy, the project coordinator.

With its results, ROADVIEW establishes important foundations for the next generation of connected and automated vehicles. The technologies and validation methods developed within the project contribute to making automated driving systems more resilient to real-world environmental conditions and support their future deployment on public roads.

About ROADVIEW

ROADVIEW (Robust Automated Driving in Extreme Weather) is a Horizon Europe innovation project focused on developing robust and cost-efficient perception and decision-making systems for automated vehicles operating in adverse weather conditions and diverse traffic scenarios. Key technology areas include adaptive sensor fusion, sensor noise filtering, collaborative perception, V2X communication, mathematically grounded sensor modeling and simulation-assisted validation. ROADVIEW brought together 15 partners from seven European countries and was supported with a total project budget of €9.7 million.