M3CAD
收藏资源简介:
M3CAD是一个为通用协作自动驾驶研究而设计的创新基准,包含204个序列,共计30k帧,涵盖了多种协作驾驶场景。每个序列包含多辆车和多种传感模式,例如LiDAR点云、RGB图像和GPS/IMU,支持多种自动驾驶任务,包括目标检测和跟踪、地图构建、运动预测、占用预测和路径规划。M3CAD的创建过程利用了Unreal Engine 5的高级渲染能力,在CARLA中模拟了真实的多车交互。该数据集支持单车和多车自动驾驶研究,旨在推动自动驾驶领域的研究。M3CAD是目前最全面的基准,专为协作多任务自动驾驶研究量身定制。数据集由德克萨斯大学北分校和丰田信息技术实验室的作者共同创建,并通过GitHub项目页面公开发布。
M3CAD is an innovative benchmark designed for general collaborative autonomous driving research, containing 204 sequences totaling 30k frames and covering a wide range of collaborative driving scenarios. Each sequence includes multiple vehicles and diverse sensing modalities, such as LiDAR point clouds, RGB images, and GPS/IMU, supporting a variety of autonomous driving tasks including object detection and tracking, map construction, motion prediction, occupancy prediction, and path planning. The development of M3CAD leverages the advanced rendering capabilities of Unreal Engine 5 to simulate realistic multi-vehicle interactions within the CARLA simulator. This dataset supports both single-vehicle and multi-vehicle autonomous driving research, aiming to advance research in the autonomous driving field. M3CAD is currently the most comprehensive benchmark tailored specifically for collaborative multi-task autonomous driving research. The dataset was co-created by researchers from the University of North Texas and the Toyota Information Technology Laboratory, and is publicly released via a GitHub project page.

- 1M3CAD: Towards Generic Cooperative Autonomous Driving Benchmark德克萨斯大学北分校, 丰田信息技术实验室 · 2025年



