Research On Automatic Assembly Technology Of Robot Based On Visual Information Fusion

In the manufacturing process,the assembly of products is essential.In order to cope with the increasingly complex assembly environment and process,the robot automatic assembly technology based on visual perception plays an important role in product assembly.The robot has the ability to perceive the external environment through visual aids.Even if the external environment changes,it can correct the relative posture in time.Building a visual measurement assistant system suitable for assembly environment for ensuring assembly precision.The main research contents are as follows:First,the development status and research trend of assembly robots at home and abroad were described.The present research status of pose estimation based on visual perception was analyzed,and the research progress of the nearest point iteration algorithm was described.In addition,the structural parameters,forward kinematics model and hand-eye calibration of the actuator are analyzed for the assembly system.Second,according to the principle of parallax distance measurement,a binocular stereo vision measurement model was established.Distortion factor was taken into account when the camera was calibrated nonlinear.A cross structure cursor positioning method was proposed for the sensor.A two-step algorithm for extracting the center line of the light strip was used to extract and fit the straight line equation in the pixel coordinate system,and used the random sampling consistency principle to fit out of the cross laser plane,3D reconstruction and error analysis of the ring gauge standard parts were finally carried out,its mean measurement error is 0.125 mm.Third,the pose estimation method of monocular camera was studied,and relevant experiments were done.An adaptive BP neural network information fusion algorithm was proposed for the measurement system,which used different numbers of sensors to carry out visual guidance assembly experiments.Finally,the pros and cons of the data obtained under different guidance methods are analyzed to prove that it is practical.Figure 35;Table 10;Reference 50...