To achieve the goals set by the European Green Deal, the European Commission plans to expand the available cycling infrastructure. As a result, the number of cyclists is expected to rise, raising safety concerns, specifically in shared spaces where multiple road users interact. One major dilemma that urban planners will face is assessing the safety and performance of cycling infrastructure. Microscopic cyclist models are capable of fulfilling this need. However, datasets for cyclist interactions in shared spaces are missing. In this work, we propose CTV-Dataset, a cyclist-focused top-view dataset collected from laboratory experiments using a camera-equipped drone. The scenarios have been specifically designed to fill the missing scenarios in literature: cyclists' lateral, bilateral, and crossing interactions with multiple road users, such as pedestrians, cyclists, and cars. Additionally, we performed other scenarios for cyclist free-flow movement and interactions with an obstructed line of sight. Lastly, motivated by the lack of suitable computer vision methods to extract trajectories from top-view videos, a general-purpose multi-object tracking tool was developed to be used for trajectories extraction. Our dataset comprises 1.68 hours of pure cyclists scenarios, with 593 clips. The dataset is available at (https://www.ifi-mec.tu-clausthal.de/ctv-dataset)
CTV-Dataset: A Shared Space Drone Dataset for Cyclist-Road User Interaction Derived from Campus Experiments
2023-09-24
1470328 byte
Conference paper
Electronic Resource
English
THE IND DATASET: A DRONE DATASET OF NATURALISTIC ROAD USER TRAJECTORIES AT GERMAN INTERSECTIONS
British Library Conference Proceedings | 2020
|