Research On Robot Automatic Grasping And Docking Technology Based On Vision

The daily maintenance of aircraft is very important to ensure the safety of flight.In addition to mechanical,electrical,visual inspection and maintenance,refueling,power supply,ventilation,water supply,refrigerant filling is also an important task.These support tasks are still more responsible by the ground staff,mechanization,automation,low degree of information.In this paper,the vision-based automatic grasping and docking technology of robot is studied,taking the unmanned aircraft support as the target scene,and the robot support of aircraft power supply,ventilation,liquid-cooled media filling and other tasks as the main research goal.Firstly,The technical scheme of automatic grasping and docking is designed.this paper analyzes the technical requirements of unmanned maintenance equipment and design the system on the basis of investigation of power supply,ventilate,refrigerant filling.Then set up the target interface visual identification and positioning,the interface automatic capture,the network remote control and monitoring,the interface visual servo automatic joint and other main functional modules.The five main hardware components of the system including industrial manipulator,control host,vision system,joint storage,docking target and more are selected and designed respectively.The design and compile the location identification detection algorithm.Target visual recognition and localization are reasonably simplified to the recognition of location mark from the actual situation of each docking target port’s relative position therefore a four circular positioning mark with distinctive visual recognition features is designed.On this basis,the dual camera acquisition and the single camera acquisition are compared and analyzed respectively.Considering the end installation requirements of the manipulator,the single camera is chosen.Focusing on the analysis and design of locating mark recognition algorithm,on the basis of image enhancement pretreatment,the target recognition methods based on image segmentation and template matching are compared and tested.And based on variable template matching,the locating mark detection algorithm is designed and compiled.Furthermore,this paper proposes a switching visual control plan combining open loop and closed loop control,analyze the coordinate transformation relationship and system calibration model of the robot grasping docking system and complete the camera calibration and hand-eye calibration of the system.During open loop control phase,with the help of the locating identification detection algorithm,the features of target feature points can be obtained.Then the location estimation algorithm realizes the goal of initial positioning and guide the robot to the location identification and the accurate target positioning can be achieved by switching the visual servo control method based on image,and the convergence efficiency is improved through adaptive gain.Then using the Rodriguez method to update the robot position.At last,Carry out software design and whole system test of software and hardware.with the robot arm as the core operating equipment ABB IRB2600 and in the context of Robo Studio6.3,RAPID programming language is used to program the control program of the Mechanical arm.Using a combination of Open CV,Halcon and other image processing libraries,in the context of MFC Visual Studio 2015,the system software module is designed,debugged,developed.After the completion of processing,testing and assembling the relevant mechanical parts,software and hardware system testing is carried out.Through debugging,the robot system can stably complete the automatic grasping and interface docking of power supply,ventilation and liquid cooling ports.The duration of one grasping operations is less than 60 seconds.The docking position error is less than 0.5mm,and the angle error is less than 1°.Therefore,it can meet the requirements of unmanned aircraft maintenance.