Framework for Quality Evaluation of Smart Roadside Infrastructure Sensors for Automated Driving Applications
This work addresses the need for quality evaluation in sensor setups for connected and automated vehicles, though it is incremental as it builds on existing datasets and metrics.
The paper tackles the problem of evaluating sensor setups for smart roadside infrastructure by proposing a multimodal framework that assesses lidar and camera sensors on accuracy, latency, and reliability using the DAIR-V2X dataset, showing it can be reliably used for future ITS-S applications.
The use of smart roadside infrastructure sensors is highly relevant for future applications of connected and automated vehicles. External sensor technology in the form of intelligent transportation system stations (ITS-Ss) can provide safety-critical real-time information about road users in the form of a digital twin. The choice of sensor setups has a major influence on the downstream function as well as the data quality. To date, there is insufficient research on which sensor setups result in which levels of ITS-S data quality. We present a novel approach to perform detailed quality assessment for smart roadside infrastructure sensors. Our framework is multimodal across different sensor types and is evaluated on the DAIR-V2X dataset. We analyze the composition of different lidar and camera sensors and assess them in terms of accuracy, latency, and reliability. The evaluations show that the framework can be used reliably for several future ITS-S applications.