四足机器人复杂地形通过数据集

采用实机户外采集 + 多源同步记录方式构建：在碎石、楼梯、斜坡、泥地等 20 种复杂地形中部署四足机器人，同步采集环境视频、机身 IMU 数据与各关节角度信息；通过时间戳严格对齐、异常数据剔除与运动状态标注，经格式标准化处理后，形成 500 小时高一致性多模态数据集，支撑地形适应与步态控制研究。

This dataset was developed via real-world outdoor data collection and multi-source synchronous recording. A quadruped robot was deployed across 20 types of complex terrains including gravel, stairs, slopes, mud and others, to synchronously collect environmental video, on-board IMU data and joint angle information of all joints. After strict timestamp alignment, outlier removal, motion state annotation and format standardization processing, a 500-hour high-consistency multimodal dataset was finally formed, which supports research on terrain adaptation and gait control.