自动驾驶数据集|SLAM数据集

<p>该数据集用于验证AVM-SLAM系统，采集于220m×110m、含430余个停车位的地下车库。包含四个1280×960分辨率的鱼眼图像序列、一个1354×1632分辨率的鸟瞰图序列（每像素对应1.05厘米，覆盖14.25m×17.18m区域）、四组轮速编码器及IMU数据，图像均以10Hz同步采集。数据集专为自动驾驶车辆在复杂地下环境中的SLAM研究设计，支持基于鸟瞰图的定位算法验证，适用于视觉-惯性-轮速融合定位方法的研发与评估。</p>

This dataset is designed to validate the AVM-SLAM system, and was acquired in an underground garage spanning 220m × 110m with over 430 parking spaces. It includes four fisheye image sequences with a resolution of 1280×960, one bird's-eye view image sequence with a resolution of 1354×1632 (each pixel corresponds to 1.05 cm, covering an area of 14.25m × 17.18m), four sets of wheel speed encoder data and IMU data. All image sequences are synchronously acquired at 10Hz. Specifically designed for SLAM research on autonomous vehicles in complex underground environments, this dataset supports the validation of bird's-eye view-based positioning algorithms, and is suitable for the development and evaluation of visual-inertial-wheel speed fusion positioning methods.