Real World No-collision Multi-sensor and V2X Road User Dataset

Overview This dataset contains 15 min 42.7 s of road-user observations recorded on 2 June 2026, from 16:44:22 to 17:00:05 WEST (15:44:22 to 16:00:05 UTC). The recorded setup comprised four cameras (3 Hanwha cameras attached to a smart lamppost and one road-placed Luxonis camera), one Radar, one LiDAR, and cooperative V2X road users. It includes CAMs from two vehicles and VAMs from one bicycle. The dataset contains parked and stationary vehicles detected by the infrastructure, including one stationary vehicle transmitting CAMs. No physical collision occurred during the recording. The release contains 97,655 infrastructure object detections and 10,499 road-user V2X messages, for a total of 108,154 records. Files Files Rows Description camera.csv 49,068 Lamppost-attached camera detections (3 cameras) represented in the CPS object format. luxonis.csv 25,851 Road-placed Luxonis camera detections represented in the CPS object format. lidar.csv 10,737 Lamppost-attached LiDAR detections represented in the CPS object format. radar.csv 11,999 Lamppost-attached Radar detections represented in the CPS object format. cam.csv 9,098 CAMs transmitted by two vehicles (one stationary). vam.csv 1,401 VAMs transmitted by one bicycle. dataset_overview.png N/A Spatial and temporal summary of all data streams. Data Origin The data was collected in Rua da Pega, Aveiro, Portugal, using infrastructure (sensors, communication devices, and vehicles) from the Aveiro Tech City Living Lab. Object detections use the application-facing objects representation from the ITS Perception Service. CAM and VAM data were decoded through Vanetza-NAP. Timestamps expressed as Unix time are in UTC. Rows are ordered by timestamp within each CSV stream.