单目视觉在仿生蜘蛛中的识别与测距应用

Recognition and Distance Measurement for Bionic Spiders Using Monocular Vision

  • 摘要: 随着仿生机器人的迅速发展及应用,关于目标信息反馈的测量方法引起了广泛关注。目前,大多数目标识别和测距方法依赖于在仿生机器人上安装雷达或双目相机等辅助装置。然而,受限于仿生机器人的资源,需要开发更轻量化、低成本且易于部署的解决方案。因此,提出了一种基于仿生蜘蛛姿态调节能力与单目相机相结合的目标检测与测距方案。该方案首先利用YOLOv5模型识别图像中的目标物体;随后,通过MATLAB对相机进行标定,获取内外参数,并对图像进行预处理、立体匹配与目标深度计算;最终得到目标物体相对于仿生蜘蛛左姿态下相机的距离。实验结果验证了该方案在目标检测与测距任务中的有效性。

     

    Abstract: With the rapid development and application of bionic robots, methods for obtaining target information feedback have attracted widespread attention. Currently, most target recognition and distance measurement approaches rely on auxiliary devices installed on the robot, such as radar or stereo cameras. However, due to the limited resources of bionic robots, there is a need for lighter, lower-cost, and easily deployable solutions. Therefore, this paper proposes a target detection and distance measurement scheme that combines the posture adjustment capability of a bionic spider with a monocular camera. In this scheme, the YOLOv5 model is first used to identify target objects in the image. Then, the camera is calibrated in MATLAB to obtain intrinsic and extrinsic parameters, followed by image preprocessing, stereo matching, and target depth computation. Finally, the distance of the target object relative to the camera mounted in the bionic spider's left posture is obtained. Experimental results demonstrate the effectiveness of this approach in target detection and distance measurement tasks.

     

/

返回文章
返回