Development Of Control Method And System Of Intelligent Sorting Robot Used In Construction Solid Waste

The construction solid waste of China has the characteristics of large total amount and rapid growth,and its recycling rate is only less than 5%.The traditional recycling process still has the problems of low recycling purity and low efficiency.This paper has developed a sorting robot for practical application scenarios.The introduction of robotic sorting in the traditional recycling process of construction solid waste can effectively replace manual sorting,avoid the dangerous factors in the manual sorting process,and improve the purity and efficiency in recycling.The work of this paper mainly includes the following aspects:1)Software and hardware platform design of sorting robot for construction solid waste.Considering various factors under actual sorting conditions,the method of assembly line operation is selected,the conveyor belt is used to transport the objects,the Cartesian robot with advantages of stable structure and strong load capacity is used as the back end.The self-designed grabbing module is installed at the end.The control system adopts the solution of industrial controller+PC,and the software of the lower computer adopts the Beckhoff Twin CAT platform and realizes the sorting functions.2)Implementation of online dynamic tracking and grasping strategy based on conveyor belt.To dynamically grasp the object on the conveyor belt,it is necessary to design a suitable algorithm to calculate the real-time position of the object,so as to plan the motion trajectory of the robot.In view of this,a dynamic grasping mathematical model is established to solve the grasping point.At the same time,in order to maximize the sorting efficiency,the dynamic object grasping strategy is designed to optimize the efficiency and recycling rate.The final experiment verifies that the average efficiency of a single unit reaches 0.974.3)Research on grasping pose algorithm based on deep learning.Aiming at the problem of low reliability when objects are in contact or overlapped in dense conditions,a grasping detection method based on deep learning is proposed.The grasping pose of the mechanical claw is expressed in a parameterized manner and a dataset is produced.After the model is trained,it can perform online prediction to output the optimal grasping of the object.The single average detection time of this model is 1.548 s,and the recall rate reaches 92.5%.The model has good generalization and detection efficiency,and also meets actual needs in the online identification process.In the end,the sorting test was carried out on the real platform,and the results showed that the success rate was basically above 90%,and the actual sorting efficiency was as close as 2000 pieces per hour.A single robot can replace the manual sorting of 3 to 5 people,which greatly improves production efficiency and reduces labor costs.