基于视觉的避障成功率检测数据集

基于视觉的避障成功率检测数据集主要面向当今自主机器人领域的迅速发展，所引起的对于视觉避障算法以及视觉避障成功率检测的高质量数据集需求。本课题基于自研的双目相机平台，完成了在动态环境复杂（人流、车流密集，路障丰富）的场景下的数据采集，主要记录相机图片观测值。

This vision-based obstacle avoidance success rate detection dataset is developed to address the surging demand for high-quality datasets for vision-based obstacle avoidance algorithms and their success rate detection, prompted by the rapid advancement of the autonomous robotics field in recent years. Built upon a self-developed binocular camera platform, this study conducted data collection in complex dynamic environments featuring dense pedestrian and vehicle traffic and abundant obstacles, primarily recording the image observations captured by the camera.

搜集汇总

背景与挑战

背景概述

该数据集是一个基于视觉的避障成功率检测数据集，主要用于自主机器人领域的视觉避障算法研究。数据集包含在动态复杂环境（如人流、车流密集场景）下采集的双目相机图片观测值，数据量为38.25GB，格式包括.zip、.bag和.rar。