首页|面向高变倍场景的变电站巡检机器人云台相机对准方法

面向高变倍场景的变电站巡检机器人云台相机对准方法

扫码查看
受导航误差和机械磨损等因素影响,变电站巡检机器人在自主巡检时的停靠位姿偏离预置位姿,导致其云台相机在高倍变焦抓图时出现目标设备不在图像内以及聚焦失败等问题.因此,提出一种面向高变倍场景的变电站巡检机器人云台相机对准方法,使巡检机器人相机能够精确对准到预置位姿,以拍摄与模板图像一致的高质量巡检图像.首先,建立巡检机器人位姿与像素误差之间的关系模型;然后,基于构造的变电站空间布局假设条件,提出机器人相机位姿误差的近似解法;最后,对相机位姿误差进行正交解耦,使用提出的折半对准控制方法校正相机位姿误差.在巡检机器人真型平台上开展相机对准试验,结果表明,相较于仅调整云台姿态的传统相机对准方法,该方法在高变倍巡检场景中表现出更高的巡检覆盖率、准确率以及更低的巡检图像像素误差.
Substation Inspection Robot PTZ Camera Alignment Method for High Zoom Scenes
Affected by factors such as navigation error and mechanical wear,the parking pose of substation inspection robot deviates from the preset pose during autonomous inspection,which causes issues such as the target equipment not in the image and focusing failure when capturing images with high zoom by using its Pan-Tilt-Zoom(PTZ)camera.Therefore,a substation inspection robot PTZ camera alignment method for high zoom scenes is proposed in this paper,which accurately aligns the inspection robot camera pose to the preset pose to capture high-quality inspection images that are consistent with the template images.First,a relationship model between pose and pixel error of the inspection robot is established.Next,an approximate solution method of the robot camera pose error is proposed based on the construction of the assumed conditions for the spatial layout of substations.Finally,the camera pose error is orthogonally decoupled and rectified by the proposed bisection control method.The camera alignment experimental results developed on the real-world inspection robot platform show that the proposed method performed higher inspection coverage,accuracy,and lower pixel errors of inspection images,compared with the traditional camera alignment methods that only adjust the PTZ pose.

inspection robotautonomous inspectioncamera alignmentpose solutionbisection control

姜骞、刘亚东、严英杰、刘庆臻、陈思、江秀臣

展开 >

上海交通大学电气工程系,上海市 闵行区 200240

巡检机器人 自主巡检 相机对准 位姿求解 折半控制

上海交通大学-联想集团校企合作研发项目

202211SJTU03-LR011

2024

中国电机工程学报
中国电机工程学会

中国电机工程学报

CSTPCD北大核心
影响因子:2.712
ISSN:0258-8013
年,卷(期):2024.44(8)
  • 23