【摘要】 协作机器人是智能制造设备的重要组成部分，“无序抓取”是提升制造业智能化水平的关键工艺，异常工况检测对于提高制造环境的安全性和机器人的操作效率也至关重要。为解决协作机器人与工件交互的异常工况检测问题，面向协作机器人的无序抓取场景，提出了一种基于电流和关节角度的异常工况检测方法。首先对协作机器人常见的异常工况进行了分类与定义。随后介绍了异常工况的分段式检测策略，通过关节角度计算度量值确定当前阶段，采用去极值滑动平均滤波和卡尔曼滤波依次对关节电流进行平滑化处理，并通过预设的关节权重因子对修正信息熵进行加权，建立阶段电流数据同协作机器人阶段工况的映射关系，再由工作周期中的所有阶段工况综合判断得到机器人总工况。最终基于UR5协作机器人设计实验验证了方法的有效性和准确性。实验结果表明，所提出的异常工况检测方法具有良好的检测效果，对所定义异常工况的综合检测准确率高达98.13%。更多还原

【Abstract】 Collaborative robots are an important part of intelligent manufacturing equipment. “Bin-picking” is a key process for enhancing the intelligence level of the manufacturing industry. Abnormal condition detection is also crucial for improving the safety of the manufacturing environment and the operational efficiency of robots. To address the issue of abnormal condition detection in the interaction between collaborative robots and workpieces, this paper proposes a method based on current and joint angles for detecting abnormal conditions in bin-picking scenarios. The paper first classifies and defines common abnormal conditions of collaborative robots. It then introduces a segmented detection strategy for abnormal conditions, determines the current stage through joint angle calculation metrics, smoothing joint currents through outlier removal sliding average filtering and Kalman filtering, weights the corrected information entropy with preset joint weighting factors, establishes a mapping relation between the stage current data and collaborative robot stage conditions, and comprehensively determines the robot’s overall condition based on all stage conditions in a work cycle. Finally, the experimental verification of the method’s effectiveness and accuracy is conducted based on the design of a UR5 collaborative robot. The experimental results demonstrate that the proposed abnormal condition detection method has excellent detection performance, with a comprehensive detection accuracy of up to 98.13% for the defined abnormal conditions.更多还原