Proof of Concept: Accelerated scenario data collection for AD verification

Purpose and goal:

The purpose of the project was to improve and evaluate the capability of Viscando traffic measurement system to provide data for scenario-based verification of autonomous vehicles (VAV).

The conclusion from the project is that Viscando data is suitable for extraction of scenarios for VAV in motorway situations, in terms of both content and accuracy.

The project also contributed to clarifying use cases and requirements for scenario data collection for VAV, while proving the efficiency of using stationary sensors for this purpose. Expected results and effects:

The capability of Viscando stationary sensors to deliver scenario data for AV verification was demonstrated through the following steps: - Motorway traffic data was collected - Accurate geometric representation of traffic objects was implemented

- Data accuracy was improved through enhanced data processing algorithms, and validated using accurate reference data collected in test vehicle - Considerable data accuracy increase was demonstrated. In parts of measurement area, the error is below 20 cm. - Future steps to achieve even better accuracy were identified

Approach and implementation:

This joint project between Viscando and Zenseact started with defining the use case for data collection and data accuracy validation metrics and methods.

Accurate reference sensor data, collected by Zenseact in a test vehicle, was used during algorithm improvements, which were run in short development sprints, and for the final accuracy validation.

Finally, the partners reviewed the the achieved accuracy, and made a common conclusion of the clear capability and strong potential of Viscando system to collect scenario data for AV verification.