Nonlinear Error Compensation And Full Closed Loop Control Of Magnetic Miniature Swimmer

Compared with traditional robots,magnetic controlled miniature soft robots have the advantages of compact structure,lightweight,large deformability,and good biocompatibility.Because the length of a robot does not exceed millimeters,it is often driven without cables.Magnetic field driving has the advantages of fast response,good penetration,and no damage to biological tissues.Therefore,the magnetic miniature soft robot is developing rapidly.In the practical application of miniature robots,error compensation and precise control are difficult problems that must be solved.For biomedical application scenarios,precise control of miniature robots is crucial.But the realization of precise servo control is hindered because of the small size,many error sources,and difficult establishment of a dynamic model of a miniature robot.Based on the above problems,this paper designs and builds a visual servo control system with the Helmholtz coil as the driving core.The system structure,driving principle,and full closed-loop control design are described in detail.On this basis,a magnetic control miniature robot with a simple structure is designed and manufactured,including its detailed structural parameters and manufacturing process,and its force and motion characteristics are analyzed.In the face of the problem that the trajectory of the miniature robot has errors and prominent nonlinear characteristics,the error sources are analyzed.It is pointed out that the maximum error comes from the error of the miniature robot itself and the error caused by the uneven distribution of the magnetic field of the Helmholtz coil,indicating that the linear compensation method cannot meet the requirements of error compensation.A method of error modeling based on deep neural networks(DNN)is proposed,which has given the parameter selection,model structure,data collection method,and data preprocessing method.On this basis,a feed-forward compensation method based on DNN is proposed to correct the forward direction of miniature robot,so as to realize nonlinear error compensation,and its effectiveness is verified by experiments.LOS-based navigation control is proposed to solve the problem of fully closed-loop control of the miniature robot.Based on the position relationship between path points and miniature robots,this method generates the forward direction in real time through vector synthesis,so that miniature robots can track the path smoothly.In order to solve the contradiction between the locomotion efficiency and tracking accuracy of miniature robot,proposed a speed control method based on fuzzy adaptive.This method makes the miniature robot faster when the path is simple(such as a straight path)and slower when the path is complex(such as large curvature and inflection point path)so that the miniature robot can still guarantee the tracking accuracy when facing complex paths.The above method is based on line-ofsight navigation control,fuzzy adaptive speed control and DNN feed-forward error compensation,which constitutes a comprehensive full closed-loop control scheme.Finally,the effectiveness of the scheme is verified by experiments.In summary,this paper proposes a comprehensive closed-loop control scheme for the nonlinear error problem in the locomotion process of the miniature robot,which effectively compensates for the nonlinear error,improves the locomotion efficiency,and realizes high-precision path tracking.It is of great significance to the practical applications of miniature robots.