Study On The Key Technology Of Redundant Glue Robot

With the rapid growth of the development of automobile production technology and production scale automobile production. The processes and technology of car manufacture are becoming increasingly complex and advanced. To ensure both the quality and quantity of the product, different kinds of industrial robots have a great debut in car production line. The industrial robot in car production line not only has a high degree of mobility, flexibility, but also a high degree of performance and intelligence. Currently, the robots with a certain redundancy are going to start to get people’s attention and favor. Compared with common industrial, the redundant robot has a wider work space and more flexible motion characteristics as a result of extra degree of freedom. The redundant robots are able to avoid obstacles and optimize the work load distribution when the end of robot is tracking the trajectory. The environment of the car production field is often full of obstacles and hard for the industrial robot to operate. How to make redundant robot automatically identify complex environment to be basis for real-time trajectory planning, and be sure to complete scheduled tasks rapidly and precisely is both the key of redundant robot technology development and where the focus of this study lies.In order to have a systematic and deep research on the above objects, the redundant glue robot which is refitted from the IRB1440 industrial manipulator is inducted as an example in the paper. The field of industrial production for specific application is as the study background of the paper. In the paper, the forward and inverse kinematics analysis, motion obstacle avoidance strategies and target reorganization of the scene which deeply influence the redundant glue working performance are studied. The specific content of the paper is as follows:Firstly, forward kinematics problem of the redundant glue robot is deduced in the paper, and the existing methods for inverse kinematics are analyzed and then a newly virtual joint method for inverse kinematics of the redundant is proposed in the paper. In order to solve the inverse kinematics of the redundant, the redundant robot is transformed to the virtual robot with only six degrees of freedom whose solving process is relatively simple. And then the solution acquired from virtual robot will be transformed to the solution of redundant glue robot. The virtual joint method is included the indicators that related to the robot joint movement restrictions and the weight function of robot motion, so the variables of all the joints of the redundant glue robot can be optimized within the limits range of joints angle. The result of the simulation shows that, when the gluing points and normal vectors are known, the angles of the robot’s joints can be solved rapidly and effectively by using the proposed method.Secondly, the trajectory planning method the redundant glue method is studied in the paper, and the workspace of the robot is solved. In order to overcome the shortcomings of the previous obstacle avoidance algorithms, an obstacle avoidance algorithm which applicable to multiple mobile obstacles is proposed in the paper. The minimum prediction distance between the obstacle and manipulator is obtained according to the state of the obstacle, and transformed to the escape velocity of the corresponding rod of the manipulator. The escape velocity is introduced to the gradient projection method to obtain the joint velocity of manipulator to complete the obstacle avoidance trajectory planning. The redundant glue robot is used in the simulation, and the result verifies the effectiveness of the algorithm.Thirdly, to achieve the identification of specific target of the robot and improve the capability of dynamic obstacle avoidance, the image segmentation algorithm based on model and profile optimization matching algorithm are studied in the paper. The inadequate of the traditional model based image segmentation algorithm is analyzed, and after that, a partial active profile model which based on original image and the difference image statistics is proposed. By the convolution of the original image and the image subtraction, the contrast of the target and background is enhanced, and then the influence of noise is reduced. The result of the test shows that, compared with other models, the proposed method in the paper has reduced the dependence on initial contour, and image with weak edges and uneven grayscale can be extracted correctly which can improve the target recognition capability of the redundant glue robot.Finally, a specific target recognition method based on bionic mechanism is proposed. In order to simplify the process of neural network training, the invariant moment is used to transform one piece of picture to a set of invariant moments with seven numbers, and acquire the invariant moment of every picture. The invariant moment related to the pictures and target number of the picture are treated as set of training sample which leads to the reduction of the samples. Additionally, the result of the simulation shows that the proposed method has shorter training time and higher recognition precision, and will promote the coating quality and work efficiency of the redundant glue robot.