为解决正交双目立体视觉自主研磨机器人对未知自由曲面的加工问题,结合全区域搜索技术、待加工轮廓的选择算法和曲面边界识别方法,提出一种适用于未知曲面加工的自主搜索方法,实现了研磨机器人对非结构化环境中未知曲面的目标搜索任务.研磨机器人利用线激光和视觉传感器来感知未知待加工曲面,采用光轴正交的两个相机同时采集研磨曲面轮廓的图像,并利用模糊逻辑方法进行局部路径规划,在此基础上通过全区域搜索技术实现对未知待加工区域的遍历.同时采用图像处理技术,即动态边界识别方法辨识目标并确定其位置.该未知待加工曲面自主搜索方法可以从曲面任意位置开始加工,算法对于非结构化自由曲面具有良好的适应性.实验结果验证了算法的可行性.%To automatically grind an unknown freeform surface by the autonomous grinding robot based on orthogonal binocular stereovision,by combining an area coverage search algorithm with a grinding surface selection algorithm for grinded profile,and surface boundary recognition method,an autonomous search strategy under unknown surface is proposed,which enables the grinding robot to fulfill target searching task in unstructured environments.The line laser and visual sensor calibration are utilized in unknown freeform surface recognition,and pictures of grinding curve are taken individually at the same time by two cameras,whose optical axes are perpendicular to each other,and a fuzzy logic algorithm is adopted for local path planning to achieve a thorough search over the concerned area.Simultaneously,an image processing algorithm referred to as dynamic boundary estimate method is proposed to recognize the target and determine its position.The designed search strategy can start from any initial position,and it exhibits exceptional adaptabilities over unstructured freeform surface.Experimental results demonstrate the performance of the proposed search algorithm.
展开▼
