Research On Posture Control Of Zebrafish Larva By Visual Servo System Based On Magnetic Tweezers

As a typical spine model organism,zebrafish larvae have received more and more attention in life science-related research in recent years.Zebrafish has become an important research object in many fields such as pathology,auxology,pharmacology,genetics,etc.for life scientists because of its high similarity with human genes,short growth cycle,strong reproductive ability,and transparent body.The body of zebrafish larva is tiny and very slender,and usually requires microscopic imaging equipment to observe its physical characteristics.In the study of zebrafish larvae,researchers often need to artificially inject specific molecular compounds,drugs,hormones and other foreign substances into the larva body,which puts extremely high requirements on the precision of the operation.At present,this operation is mainly done by manual injection by experienced experimenters.However,different experimenters will cause a series of problems such as poor consistency,low success rate,and high larva damage during the operation,which will seriously affect the follow-up life science research.Although the existing microinjection equipment for zebrafish larva can achieve basic injection functions,such equipment mainly uses microfluidic technology and negative pressure holding to control the position and posture of larva,which cannot ensure high-throughput injection while achieving nondestructive capture of the larva fish.Although the traditional technology of cellembryo-oriented microinjection system has matured and commercialized,it is still difficult to achieve effective posture control for zebrafish larvae with irregular shapes,autonomous consciousness,and complex internal structure.This article focuses on the problem of zebrafish larva posture control in microinjection,and conducts in-depth and innovative research,proposing a set of design solutions for the zebrafish larvae visual servo posture control system based on magnetic tweezers.The main content of this article can be summarized as:To solve the problem of visual feedback in the micromanipulation of zebrafish larva,based on the traditional computer vision algorithm,it is proposed to use a new solution consisting of mechine learning algorithms such as K-means clustering,Gaussian Mixed Model(GMM),Expectation-Maximization algorithm(EM),and Principal Component Analysis(PCA)to realize the functions of multi-target juvenile fish identification,position and posture judgment,target tracking,etc.in the microscopic field of view.On the basis of ensuring the speed of image processing,the accuracy and robustness of the image recognition algorithm have been greatly improved,which has laid the foundation for subsequent tracking and attitude control of larva fish.About the trajectory planning problem in the micromanipulation of zebrafish larva,this paper analyzes the current mainstream trajectory planning algorithm and proposes a new zebrafish larva trajectory planning scheme based on the artificial potential field method.On the basis of extracting the position and body orientation of many larvae in the field of vision through image algorithms,a set of potential energy function design schemes are designed,and a variety of deep learning optimization algorithms are used to separately plan for each juvenile from its current location.The movement trajectory from the current position to the target injection area provides guidance for the subsequent manipulation of the magnetically controlled robot.In view of the problem of the platform construction of the magnetic control system,this paper analyzes the control characteristics of the two magnetic control coils in detail,designs two sets of control schemes according to the different magnetic field characteristics,and completes the experimental verification of one of the schemes.Based on the platform construction,an advanced control algorithm based on PID algorithm is designed for the existing device,which can basically achieve stable control of the magnetic tweezers.In this paper,the proposed design scheme,except for the magnetic tweezers control scheme based on rotating magnetic field,has been experimentally verified.On the basis of simulation and experimental results,the important indicators of system performance(stability,accuracy,etc.)are analyzed in detail,and improvement schemes are given.