| “The catalog of the key point research and development of coal mine robot”requires that a drilling robot for water prevention,for outburst prevention and scour prevention has the functions such as automatic handling of drill pipes,automatic adjustment of drill rig attitude,intelligent drilling plane and positioning.The existing automatic drilling rig with drilling pipe loading and unloading function,its drilling pipe loading and unloading device mechanism is complex,multi-level transmission,no intelligent decision,high failure rate,low grasping efficiency.At the same time,the existing drilling rig position and pose adjustment depends on manual adjustment,with low degree of automation and poor accuracy.The position and pose adjustment mechanism cannot compensate for errors due to wear and deformation after long-term use,which makes it impossible to achieve accurate extraction.According to the requirements of drilling robot construction hole group in the whole process,in the whole process of construction,the drilling robot can automatically change the position of the next drilling position by calling the pose adjustment module;The automatic rod identification module is then used to intelligently identify the new drill pipe delivery location.The automatic rod identification module and the precise position and pose adjustment module cooperate and cooperate with each other,and become the key to restrict the construction efficiency of the whole process.Therefore,the key technology research of automatic rod adding recognition and drilling arm positioning error compensation can greatly improve the construction efficiency of drilling robot.At the same time,the application of this technology can also promote the process of "mechanized replacement,automatic reduction,intelligent unmanned",and has important theoretical significance and engineering application value.By means of the analysis of the application status of the coal mine underground drilling robots,the key technical problems restricting the development from the automatic drill rig to the intelligentized drilling robot were clarified.Aiming at the roadway conditions in underground limited environments,the overall plan for two-component walking drilling rig was proposed including the pipe handling system and drilling robot system was put forward,In combination with the demand of the 3D visual recognition and the monocular vision recognition of the automatic drill pipe handling,the set-up and the structure design of the drill pipe-grabbing subsystem and the drill pipe-delivering subsystem were completed,providing the subsequent research with the support conditions.Aiming at the status of grabbing failure due to wearing of the drill pipes when the mechanical arm with 6 degrees of freedom stacks up the drill pipes,the structural3 D vision recognition plan was proposed.Through the remodel design and the kinematics analysis of the explosion-proof mechanical arm,and hand-eye calibration of the mechanical arm and 3D explosion-proof camera,the basis for the recognition and positioning of the drill pipe-grabbing subsystem was provided.The point cloud was captured by 3D explosion-proof camera.Segmentation,denoising,down-sampling,feature detection of the point cloud,normal recognition of the drill pipes and histogram of the point features were made,and the accurate location information of the drill pipes was gotten from the rough and precise matching of the point cloud of the drill pipes.In consideration of the influence of the complicated conditions such as the downhole fallen coal and rocks,parasitic light,dust and spray,the vision recognition method merging the 3D vision and the stacking algorithm was proposed,which can recognize accurately and robustly the drill pipes in a large-volume drill pipe cabin.Aiming at the feature of the flexible layout of the tow-component drilling robot,the monocular vision recognition based on the multiple collaboration targets was adopted,solving the problem of recognizing the delivering position of drill pipes under conditions of long distance,large visual angle and being easily shield.In order to meet the precision requirements of the delivering position at radial direction,axial direction,angle of pitch and yaw angle,based on the results of the internal reference calibration and the hand-eye calibration,combined with extraction and recognition of the features of the collaboration targets,P3 P algorithm was adopted to calculate the information of the delivering position of the drill pipes to realize the accurate acquisition of the drill pipe-delivering position.In view of the complicated environments such as underground backwater,return slag,dust,spray and direct glare of miner’s lamp,the proposed multi image-enhancing algorithm increased the reliability of image recognition.The laser projection method was used to carry out the precision test of the large-angle position and attitude,the monocular vision error compensation method for the error curve-fitting method the kinematics reckoning algorithm was put forward.The test proved that it can solve the problem of out-oftolerance of the large angle.By analyzing the structural action of the frill boom,a kinematics model of thedrill boom was set up.The causes of the position and attitude error of the drill boomwas deeply analyzed.According to the perturbation differential theory and theflexibility analysis method,the static and dynamic error compensation models werebuilt.Based on RBF neural network method,the reverse solution of thecompensation model of the drill boom was calculated,the joint variable obtainedfrom the reverse solution controls the accurate positioning of the drill boom.Aimingat the non-real time at static and dynamic error compensation measurement,amethod for detection of position and attitude error by laser range finder andhigh-precision trepanning directional device was proposed.The error compensationmodel was set up.The on-line error detection and compensation was realized.Thetest results indicated that the compensation precision met the requirements.And thetraditional error detection methods were used to verify the correctness of the errorcompensation model of the position and attitude by the combination of the laserranger.The working environments in underground coal mines were simulated in the laboratory to test the recognition effect of 3D and monocular vision at simulation of exceptional conditions such as dust,spray,strong light exposure and external disturbance.The regulating process of the underground rig and trepanning positioning was simulated to test the leveling and positioning error.The test results demonstrated that it met the precision requirements.The full operation process was simulated to test the full process precision and the reliability.The results indicated: drill pipe-grabbing and delivering accuracy rate were 100% and 99.8% respectively;the drilling robot was used to drill automatically 6 boreholes in Pingbao Coal Industry Co.,Ltd,Henan Province,totally 604 m,drill pipe-grabbing and delivering accuracy rate was100%,verifying the precision and the reliability of 3D vision algorithm and stacking algorithm,recognition algorithm of the collaboration targets in complicated scene and drill boom positioning error compensation algorithm. |
PDF Full Text Request