In collaboration with Teede Tehnokeskus and TalTech, a 3D mapping and machine learning solution will provide comprehensive and integrated street marking detection system. Our StreetVision Pro solution automates monitoring activities, reducing the need for human resources and allowing city personnel to focus on other critical tasks. New data will be collected and calibrated against previously gathered data to ensure accuracy and reliability. Existing data from Teede Tehnokeskus and the cities, such as information on road markings and other urban objects, will be used to train AI models, improving their precision and enabling more accurate analysis. TalTech will apply their expertise in multi-resolution and real-time computer vision to streamline and optimize data quality to improve AI reliability and reduce data processing effort. City staff will be involved as key stakeholders throughout the development, training, testing and validation to ensure that the solution provides an optimal fit for their long-term needs.
By integrating both aerial data collected from Unmanned Aerial Vehicles (UAVs) and ground-based data collection, the solution will provide a complete overview, covering areas inaccessible to UAVs, such as tunnels, underpasses, and areas obscured by trees. This combined approach ensures full coverage and overcomes the limitations of using UAVs alone. Additionally, the solution will seamlessly integrate with existing systems, like Tartu’s ArcGIS-based work order platform, to automate the creation and management of maintenance tasks. Key use cases include monitoring road markings and generating automated work orders to maintain standards. The solution’s flexibility also allows it to support various other city management tasks, such as inventory of poles, traffic signs, landscaping, gauges, slopes and water accumulation, facilities and other elements located in the road environment. The three-dimensional model of the road space created by the software can also be used by self-driving vehicles for orientation on the road.
The proposed solution includes point clouds, orthophotos, automatic road marking detection, and automated work orders. Additionally, if there is client interest, we are developing an additional module that allows clients to perform quick
measurements of specific sections between larger surveys (1-2 times a year) and upload the data into the system as needed.
The solution builds on the existing 3D mapping solution of Teede Tehnokeskus, which already includes a machine learning component that will be significantly further developed, especially in data analysis, and expanded with new capabilities to analyze data from UAVs. As a result, a unified system and environment will be created, ensuring that data processing is fast enough to keep road markings in the cities consistently compliant with standards and of high quality.
