Record & Recognize the Road Environment with Sensorics & AI

The CDV – Transport Research Centre faced the challenge of modernizing its measurement vehicle. The outdated cameras and user interface no longer met the strict requirements for regular road network asset management and the process of recording and processing data was simply too slow.

Our team identified sensors that adapted the vehicle and expanded their functionality. In addition, we created an automation that successfully reduced the need for manual data processing during asset recording. Thanks to our neural network the result is a blend of innovation and practicality, enhancing overall efficiency in operations while lowering personnel requirements.

Goals:

• Modernize the measurement vehicle for road asset inventory purposes,
• integrate state-of-the-art sensors to provide better visibility of vehicle surroundings: LiDAR, camera + stereo camera, localization units, radars,
• accelerate data post-processing through software automation,
• add an advanced functionality of detection and classification of traffic signs.

Challenges:

• Identification of suitable sensors providing sufficient image quality,
• training robust neural networks for detecting and classifying traffic sign locations (front, rear) and multi-level identification of signs shape/type/detail (e.g. circular/speed limit/80 km/h).

“Thanks to the new equipment in our measurement vehicle and the integration of key functions with innovative software solutions, we can address the challenges of science and research, such as ensuring migration routes for flying vertebrates. The use of the measurement vehicle also has practical applications, for example, in enhancing the safety of roadways or in traffic safety assessments of the existing road network.”

― Pavel Havránek, Director of the Division of Traffic Engineering, Safety, and Strategies at CDV

Solution:

Based on our partner’s needs, our team first focused on a thorough selection of advanced sensors that we implemented in the measurement vehicle including a  LiDAR and localization unit. This enabled higher accuracy GPS position recognition and also provided a 3D perspective of objects, which plays a key role in recording the road environment.  At the same time, we developed sophisticated software that automated the operation of the sensors. This increased measurement efficiency, particularly by reducing the need for manual data processing, allowing these tasks to be handled by one operator instead of two.

In the next phase, based on feedback from the Transport Research Centre, we integrated additional functionality that enables automatic detection and recognition of vertical traffic signs. This feature simplifies the asset inventory process and helps update the national catalogue of test areas for autonomous vehicles, which The Transport Research Centre created. Thanks to these new features, a database of information is available for each traffic sign, containing its location, photograph, and classification.

“Our ongoing collaboration with Roboauto not only brings new insights but also practical solutions that help improve user trust and safety in automated transportation. In this way, we support innovations that have the potential to significantly contribute to the development of sustainable mobility and enhance the efficiency of transportation systems. Together, we are working to ensure that the future of transportation is not only technologically advanced but also accessible and safe for all users.”

― Pavel Havránek, Director of the Division of Traffic Engineering, Safety, and Strategies at CDV

If you are looking for a reliable partner for the development of innovative technologies that advance transportation, contact us at info@roboauto.tech.