Design Of Curbstone Automatic Paving Mechanical System Based On Monocular Vision

As the urbanization process is getting faster and faster,the demand for road construction is increasing.At present,the pavement of shaped curbstones is mainly done manually.The manual paving method has the disadvantages of time-consuming,labor-intensive and low efficiency,which affects the road construction progress.Therefore,it is imperative to improve the curbstone paving efficiency in road construction.The current visual recognition technology is developing rapidly in all walks of life.This subject is based on monocular vision,designing a curbstone automatic pavement mechanical system to improve the curbstone paving efficiency,focusing on the identification and positioning of molding curbstones and robot trajectory planning.The main research contents are as follows:(1)Design on structure of curbstone paver.The structural layout of the curbstone paving machine is carried out,focusing on the design of the curbstone transportation system and the paving mechanism,and the workflow of the curbstone paving machine is given.(2)Design on visual robot control system.Firstly the entire control system plan is designed,the control system includes curbstone identification and positioning system and curbstone grasping system,then the hardware and software of the system are selected,and finally the human-computer interaction interface in the Visual Studio environment is developed.(3)Design on curbstone identification and positioning algorithm.Aiming at the geometric features of the curbstone-shaped rule and the surrounding environment where the curbstone is located,a corresponding identification and positioning algorithm is designed.The algorithm aims to complete tasks such as camera calibration,real-time acquisition,image processing,and template matching,and finally to accurately locate the curbstone center of mass.(4)Robot kinematics analysis and trajectory planning.The D-H parameter model of the IRB6620 robot is established,and the forward and inverse kinematics solutions of the robot are derived by formulas.Then the trajectory planning of the robot is carried out under the joint space,and the kinematics simulation and trajectory planning are carried out with Matlab software.(5)A curbstone model recognition and grab experiment system based on monocular vision was established.The centroid coordinates of the curbstone model obtained in Halcon are transmitted to the controller of the robot,and the controller drives the robot to grab the curbstone model along the planned smooth trajectory.Experiments show that the recognition and positioning system can accurately locate the center of mass of the curbstone model,and the robot can complete the grabbing task of the curbstone model.