Tsukada Laboratory (T-Lab), The University of Tokyo, is pleased to announce the public release of the URT-Kashiwa Dataset, a novel driving dataset designed to support research on unprotected right turns, one of the most challenging maneuvers for autonomous vehicles.
About the Dataset
The URT-Kashiwa Dataset (Unprotected Right Turn, Kashiwa Campus) provides high-quality recordings of driving scenarios specifically focused on unprotected right turns—an essential case study for understanding decision-making in urban traffic. These scenarios are widely recognized as among the most difficult for autonomous driving systems to handle, given the need to negotiate with both oncoming vehicles and pedestrians. While unprotected left turns are a major focus in European traffic contexts, in Japan this challenge is represented by unprotected right turns.
Data Collection
The dataset was collected at the Kashiwa Campus of the University of Tokyo using the Pixkit 3.0 autonomous vehicle platform. The recordings include controlled interactions with:
- One pedestrian
- One opponent vehicle
- 16 different routes
Access
The URT-Kashiwa Dataset is available to the research community. For access instructions and technical documentation, please visit:
👉 [Link to dataset download / project page]
License
This dataset is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0).
