RoboSense
收藏资源简介:
RoboSense数据集由上海交通大学和SenseAuto研究共同创建,专注于低速自动驾驶车辆的近场场景理解。该数据集包含超过133,000帧同步数据,覆盖7,600多个时间序列,标注了140万个3D边界框和轨迹ID。数据集通过多种传感器(摄像头、激光雷达和鱼眼镜头)收集,确保了全方位的视角覆盖。创建过程中,数据集在多种场景下进行了采集和标注,特别关注了近距离障碍物的检测和跟踪。RoboSense数据集的应用领域主要集中在自动驾驶技术的研究,特别是在低速环境下的障碍物感知和预测任务。
RoboSense Dataset was co-created by Shanghai Jiao Tong University and SenseAuto Research, focusing on near-field scene understanding for low-speed autonomous vehicles. This dataset contains over 133,000 frames of synchronized data, covering more than 7,600 time series, with 1.4 million annotated 3D bounding boxes and track IDs. It is collected via multiple sensors including cameras, LiDARs, and fisheye lenses, ensuring comprehensive coverage of perspectives. During its creation, the dataset was collected and annotated across various scenarios, with particular emphasis on the detection and tracking of close-range obstacles. The RoboSense Dataset is primarily applied to autonomous driving technology research, especially for obstacle perception and prediction tasks in low-speed environments.

- 1RoboSense: Large-scale Dataset and Benchmark for Multi-sensor Low-speed Autonomous Driving上海交通大学计算机科学与工程系,SenseAuto研究 · 2024年



