基于工业机器人动力学模型的差分力矩偏差碰撞检测方法  

Differential torque deviation collision detection method based on the dynamic model of industrial robot

作　　者：张振宇[1,2] 钟佩思 张超[1,2] 王祥文[1,2] 刘梅[2] 孙钰雁 ZHANG Zhenyu;ZHONG Peisi;ZHANG Chao;WANG Xiangwen;LIU Mei;SUN Yuyan(Advanced Manufacturing Technology Research Center,Shandong University of Science and Technology,Qingdao 266590,CHN;School of Mechanical and Electronic Engineering,Shandong University of Science and Technology,Qingdao 266590,CHN;Qingdao Hisense Hitachi Air Conditioning Systems Co.,Ltd.,Qingdao 266510,CHN)

机构地区：[1]山东科技大学先进制造技术研究中心,山东青岛266590 [2]山东科技大学机械电子工程学院,山东青岛266590 [3]青岛海信日立空调系统有限公司,山东青岛266510

出　　处：《制造技术与机床》2023年第3期163-172,共10页Manufacturing Technology & Machine Tool

基　　金：山东省自然科学基金资助项目(ZR202103070107,ZR2020MF101)。

摘　　要：针对工业机器人的碰撞检测问题,提出一种基于工业机器人动力学模型的差分力矩偏差碰撞检测方法。采用基于惩罚函数的自适应遗传算法计算动力学参数辨识中的最优激励轨迹;采用自适应学习率的单层神经网络实现机器人动力学参数辨识,得到机器人动力学模型;利用该动力学模型和工业机器人自身携带传感器及驱动电流信号等计算预测力矩和实际力矩的力矩偏差,对力矩偏差进行向后一阶差分和向后二阶差分,从而将工业机器人碰撞区分为有意碰撞和无意碰撞,实现机器人碰撞检测并触发相应的避撞机制,避免造成设备及人员损坏。用6R工业机器人进行实物碰撞检测验证,实验结果表明碰撞检测算法能够有效区分无碰撞、无意碰撞和有意碰撞,提高了工业机器人协同作业的安全性。Aiming at the problem of industrial robot collision detection, a differential torque deviation collision detection method based on the dynamic model of industrial robots is proposed. An adaptive genetic algorithm based on penalty function is used to calculate the optimal excitation trajectory in the identification of dynamic parameters. A single-layer neural network with an adaptive learning rate is used to identify the dynamic parameters of the robot, and the robot dynamic model is obtained. The torque deviation between the predicted torque and the actual torque is calculated by using the dynamic model, the sensor carried by the robot and the driving current signal.The backward first-order difference and backward second-order difference of the torque deviation are carried out to distinguish the industrial robot collision into intentional collision and unintentional collision, realize the robot collision detection, and trigger the corresponding collision avoidance mechanism to avoid equipment and personnel damage. Physical collision detection is verified with 6R industrial robots, and the experimental results show that the collision detection algorithm can effectively distinguish noncollision, unintentional collision and intentional collision, which improves the safety of collaborative operation of industrial robots.

关 键 词：碰撞检测 参数辨识 力矩偏差 差分 

分 类 号：TP241[自动化与计算机技术—检测技术与自动化装置]