Research On Coordinated Control Method Of Dual-arm Robot For Surgical Instrument Sorting Tasks

In the context of the wide application of robotics technology in the various walks of life,China vigorously promotes the development of medical robots.At present,the sorting tasks of surgical instrument in large hospitals are usually completed by medical staff by hand,and its research in the field of robotics is still in the blank.Due to the large usage and special structure of surgical instruments,the manual processing method has several drawbacks: it is time-consuming,poses an infection risk and has potential for sorting errors.In order to complete the sorting task accurately and avoid these threats,a sorting robot system based on dual-arm coordination is built in this paper,which can realize various sorting operations and experimental tests of surgical instruments under the guidance of visual information.This paper can be divided into three parts: the construction of the sorting robot system for surgical instruments,the dual-arm coordinated compliance control strategy,and the extraction and detection of visual information of target instruments.A sorting task planning for dual-arm robots is designed to simulate the actual sorting process of surgical instruments in hospitals.In order to accomplish the operation of lifting,grasping,unlocking,opening and placing of surgical instruments under the condition of dual-arm coordination,the mechanical system of the sorting robot based on dual-arm coordination is built and its software control platform is designed.The model and analysis of the two-arm robot system are carried out.Considering that the coordinated operation process of a dual-arm robot may produce a great force because of the position error,the constrained coordinate system and the constraint relationship of the coordinated motion are established to prevent the damage between the robot and the manipulated object.Three kinematic models of dual-arm tightly coordinated operations are analyzed.A hybrid fuzzy control strategy is proposed to adjust the contact force adaptively during the unlocking operation of surgical scissor.A comparative experiment of three different control strategies is performed to validate its effectiveness.In order to effectively clamp the target instruments which is randomly placed,binocular stereo vision is used to reconstruct the three-dimensional image of the surgical instruments.Template matching and matrix transformation are used to identify and locate the target instruments respectively.A pose estimation method for surgical scissors based on texture analysis is proposed.To distinguish whether the surgical instruments are contaminated with blood or not,the data sets of blood-contaminated instruments and non-blood-contaminated instruments are established.A deep neural network model is presented to detect blood traces of post-operative instruments,which provides important evaluation indicators for instrument classification.