WiDEVIEW

WiDEVIEW是由德克萨斯A&M大学机械工程系创建的首个多模态数据集，专注于城市自动驾驶场景中的车辆与行人交互。该数据集整合了激光雷达、三个RGB相机、GPS/IMU和超宽带传感器，共包含21个序列的传感器数据，旨在通过高精度的传感器融合技术，提升对复杂交通场景中行人及车辆行为的理解和预测。数据集的创建过程涉及在校园及繁忙道路上的实际驾驶数据采集，以及对图像数据的2D边界框标注。WiDEVIEW的应用领域主要集中在提升自动驾驶车辆的安全性，特别是在处理与行人的交互时，通过增强超宽带传感器的使用，解决视觉系统在特定条件下的局限性。

WiDEVIEW is the first multimodal dataset developed by the Department of Mechanical Engineering at Texas A&M University, focusing on vehicle-pedestrian interaction in urban autonomous driving scenarios. This dataset integrates LiDAR, three RGB cameras, GPS/IMU, and Ultra-Wideband (UWB) sensors, containing a total of 21 sequences of sensor data. It aims to improve the understanding and prediction of pedestrian and vehicle behaviors in complex traffic scenarios through high-precision sensor fusion technologies. The dataset creation process involves real-world driving data collection on campus and busy arterial roads, as well as 2D bounding box annotation for image data. The primary application scope of WiDEVIEW centers on enhancing the safety of autonomous vehicles, particularly when handling pedestrian-vehicle interactions, by leveraging enhanced UWB sensor deployment to address the limitations of visual systems under specific conditions.

搜集汇总

背景与挑战

背景概述

WiDEVIEW是一个多模态城市自动驾驶数据集，整合了多种传感器数据，专注于车辆与行人交互，旨在提升自动驾驶车辆的安全性。数据集包含21个序列的传感器数据和2D边界框标注，特别强调通过超宽带传感器增强对行人行为的理解和预测。