输电线维护机器人用于代替人工完成危险作业,准确的故障检测与合理的行为规划对于作业效果至关重要.针对以上需求,采用视觉方法,提出了一种基于图像特征分类的输电线断股检测方法.该方法提取边缘梯度向量作为图像特征,采用支持向量机方法进行分类运算完成线路断股检测.在断股检测的基础上,利用断股检测信息与机器人传感器测得的信息构建机器人状态向量.根据当前状态向量,结合机器人断股补修作业流程,提出了面向捋线与压接复杂作业的机器人断股补修作业行为规划方法.利用实验室模拟线路开展实验,验证了提出的输电线断股检测及行为规划方法的有效性.
Power line maintenance robots are used to replace workers due to the dangerous maintenance operation, and the robot maintenance effect is much related with accurate fault detection and rational behavior planning. With those requirements in mind, a visual method is presented to detect the broken strand fault based on the classification of an image feature. In the visual detection method, image edge gradient histogram is firstly extracted as the image feature, and broken strand detection can be accomplished by the classification of the image feature with support vector machine (SVM) method. On this basis, several robot state vectors are established by combining the broken strand detection result and the information of robot sensors. Based on the current state vector and robotic broken strand repair process, a behavior planning method for broken strand repair is proposed toward complex operations of broken strand return and clamps installation. Experiments are carried out in the laboratory, and results demonstrate the effectiveness of the proposed broken strand detection method and the behavior planning method.