Research And Development Of Key Technologies Of Intelligent Palletizing Robot

Nowadays, the application of robot is more and more popular. Especially with the development of manufacturing and logistics industry, palletizing robot technology has attracted more and more attention. However, the domestic robot technology research and application of the intelligent degree is low, and the relevant technology is not mature enough. One of the reasons for these problems is the impact of foreign robot market and another reason is the weak strength of domestic related technology research. In order to improve the intelligent level of the robot palletizing, the key technologies such as the visual system of the palletizing robot multi sensor information fusion are researched in this paper.The research progress and application situation of Intelligent Palletizing Robot in domestic and overseas is analyzed.The application characteristics of intelligent palletizing robot overseas and deficiencies on research and application of the domestic ones are concluded, the key points of this paper is put forward. On the basis of the requests of the test and study, the proper camera, lens and image acquisition card are selected, thus forming binocular vision system of this paper. The visual system coordinate system and camera imaging model and the visual model are analyzed, the measuring principle of binocular vision system and binocular vision calibration method are studied. Using MATLAB toolbox and a graphic dot target calibration method respectively the calibration of the binocular vision system, contrast calibration precision of the two methods, and the higher precision calibration method is selected.The mathematical model of the binocular vision measurement is analyzed, and the vision system of coordinates of binocular method is obtained. In order to obtain high measurement precision, establishing the accuracy of binocular vision measurement model, the influences of focal length and the angle between the optical axis and the line and other factors on measurement accuracy are analyzed based on the geometric method, and the methods of reducing errors is got. To improve the matching point positioning accuracy, the edge points fitting straight line intersection point positioning method of measuring point. The inverse kinematics solution, positive solution, work space and path planning of ABB IRB2400 are resolved. On the basis of the test requests of this paper, the proper sensor is selected, thus building up multi-sensor information fusion palletizing object recognition system. Comparing thecharacteristics of multi-sensor information fusion method, BP neural network is used for the fusion of visual, force and thermal sensors information to identify the stack object.The vision system of palletizing robot and grab experimental platform are built.Experiments are carried out to verify the theoretical methods of visual measurement,path planning and multi sensor information fusion. The experimental results prove that binocular measuring error is 0.87%~2.51%, and the judgment accuracy rate of the sensor fusion system can reach 91.7%. The feasibility of this method is verified by experimental results.