扫地机器人L型墙+小物品场景测试数据

扫地机器人在工作时时常会遇到一些特殊场景，比如L型墙场景且有小物品存在的场景，基于视觉的扫地机在此场景中获取的深度数据常有边界不清晰、无法识别细小物体等组合复杂问题，如何能在此场景中准确定位与避障是个难点也是重点，在此场景，扫地机深度视觉模组获取的深度数据的精度就显得十分重要。搭建扫地机真值系统，用真值相机和目标深度模组进行不同材质、光线、距离下L型墙+小物品场景的数据采集，获取L型墙+小物品场景的数据，并对数据进行处理得到校正后的数据用于目标深度模组的校正，提升深度模组在扫地机L型墙+小物品场景下的测量精度，以解决扫地机定位与避障问题。采扫地机器人真值系统采集不同材质、光线、距离下的L型墙+小物品场景的数据，包括原始深度相机数据、真值相机采集的点云数据；对原始深度相机数据进行系统误差去除；对真值相机采集的L型墙+小物品场景点云数据进行修复，通过平面拟合算法进行L型墙+小物品场景的墙面校正、再通过滤波算法对点云进行平滑去噪，得到修复后的真值相机点云；将修复后的真值相机点云与深度相机进行点云配准，得到校正后的L型墙+小物品场景的点云数据——校正后数据laser-x、laser-y、laser-z和校正后数据tof-x、tof-y、tof-z。

When operating, sweeping robots frequently encounter special scenarios, such as L-shaped wall scenarios with small objects. Vision-based sweeping robots typically face complex combined issues including blurry boundaries and failure to recognize small objects in the depth data captured in such scenarios. Achieving accurate positioning and obstacle avoidance in these scenarios is both a critical challenge and a key research focus, making the accuracy of depth data acquired by the depth vision module of sweeping robots extremely important in such cases. We built a sweeping robot ground truth system, which uses a ground truth camera and a target depth module to collect data of L-shaped wall + small object scenarios under different materials, lighting conditions and distances. The collected data of L-shaped wall + small object scenarios are processed to obtain calibrated data for calibrating the target depth module, so as to improve the measurement accuracy of the depth module in L-shaped wall + small object scenarios for sweeping robots, and solve the positioning and obstacle avoidance problems of sweeping robots. The sweeping robot ground truth system collects data of L-shaped wall + small object scenarios under different materials, lighting conditions and distances, including raw depth camera data and point cloud data collected by the ground truth camera. Systematic errors are removed from the raw depth camera data. The point cloud data of L-shaped wall + small object scenarios collected by the ground truth camera are repaired: first, wall surface calibration for the L-shaped wall + small object scenarios is performed via plane fitting algorithm, then smoothing and denoising are conducted on the point cloud via filtering algorithm to obtain the repaired ground truth camera point cloud. The repaired ground truth camera point cloud is registered with the depth camera's point cloud to obtain calibrated point cloud data of L-shaped wall + small object scenarios — calibrated data: laser-x, laser-y, laser-z and calibrated data: tof-x, tof-y, tof-z.

搜集汇总

数据集介绍

特点

该数据集为扫地机器人在L型墙和小物品场景下的测试数据，包含1122条记录，每年更新一次。数据涵盖了模组安装参数、采集环境信息及原始与校正后的深度数据，旨在提升扫地机器人在复杂环境中的定位与避障能力。