CARLO
收藏arXiv2025-09-30 收录
下载链接:
https://github.com/Stanford-ILIAD/CARLO
下载链接
链接失效反馈官方服务:
资源简介:
该数据集名为CARLO,是一款家庭用2D驾驶模拟器,它以简化的方式模拟物理特性并处理视觉呈现,这使得可以开发控制模型并收集大量数据。CARLO模拟了车辆、建筑物和行人,采用点质量动力学模型。每个模式收集的数据量达到了80000个剧集。该模拟器的任务是针对接近事故场景进行驾驶模拟和数据收集。
The dataset is named CARLO, a home-use 2D driving simulator that simplifies physical simulation and visual rendering, enabling the development of control models and the collection of large-scale datasets. CARLO simulates vehicles, buildings and pedestrians, employing a point-mass dynamics model. Each mode yields up to 80,000 episodes of data. The simulator is tailored for driving simulation and data collection in near-accident scenarios.
提供机构:
Stanford-ILIAD



