Research On Perception And Control Of Full-automatic Venipuncture Robot

Venous puncture is a routine method of diagnosis and treatment in hospitals,which is widely used in venous blood collection,blood transfusion,tube placement and drip.Venipuncture is usually done manually by medical staff,and the puncture failure rate is high,so the failure rate is high.Especially in cases such as COVID-19,it is easy to cause cross infection of medical staff.Therefore,it is necessary to use robots instead of medical staff to perform venipuncture.Based on the principle prototype of the first generation venipuncture robot,this thesis completes the design of the control system software and hardware,designs the vein perception positioning algorithm and the control strategy combining visual guidance and force feedback,to realize the robot’s perception and control.First,by analyzing the process of medical staff performing venous blood collection,the process of the robot performing venipuncture is designed,and the design index of the robot is proposed based on the characteristics of the human vein.Based on the first-generation principle prototype,a multi-layer control system is designed,and the hardware selection of the perception and decision layer and the ontology control layer is completed respectively.Based on the hardware,a modular design method is used to design the software system architecture.Then,on the basis of the existing perception and decision-making hardware,the vein perception and positioning algorithm is designed and implemented.The algorithm is divided into two parts: the first part is vein image processing to realize the positioning of the vein in the two-dimensional image,the algorithm is divided into the steps of filtering noise reduction,image enhancement,threshold segmentation,morphology processing and skeleton extraction,the second part is stereo Matching algorithm,this thesis draws on the cross-scale cost aggregation framework,designs and implements a stereo matching algorithm based on cross-scale cost aggregation,uses Gaussian downsampling to obtain image sets at different resolutions,and performs cost calculation and cost aggregation at different scales.Under the premise of ensuring accuracy,the speed of the algorithm is greatly improved.After that,robot body control is performed,including trajectory planning based on genetic algorithm and impedance control based on puncture force feedback.Trajectory planning implements the selection of a pre-position that is easy to puncture according to the puncture decision,and then uses genetic algorithm to optimize the trajectory between the zero position and the prepuncture position to obtain the optimal time planning.In the robot control method,this thesis introduces impedance control based on puncture force feedback to achieve a control strategy that combines visual guidance and force feedback control.Finally,this thesis designs the experiment and simulation.In the experiment,the light source conditions were changed in terms of hardware,different image enhancement algorithms were used in the algorithm,and the image processing results were compared.The simulation includes control strategy simulation and soft tissue puncture simulation.Control strategy simulation is used to verify the impedance control based on puncture force feedback,and to study the control effect of the controller under different impedance parameters.Soft tissue puncture simulation is used to simulate the puncture process of human tissue,study the puncture force during the puncture process,and study the change of the puncture force under different needle insertion angles,positions and speeds.