| Microscopic manipulation is a technique for processing biological cells or embryos at the microscopic scale,which provides an entirely new way for humans to understand and modify the natural world.Among them,the capture technique of biological objects at microscopic scale is one of the important research contents of microscopic manipulation,which has received wide attention from researchers at home and abroad in recent years.As biological cells and tissues are fragile,prolonged exposure to air or excessive mechanical contact with manipulators may cause irreversible damage to biological objects.Therefore,the operation process should meet the premise of high precision and high specificity and should fully consider the rapidity and low damage of operation.Based on the above problems,this paper develops an automated,fast and low-damage biological capture system,and conducts in-depth research around visual perception and motion control.The main work is as follows:First,the capture system actuators and biological target localization strategies are investigated.A particle swarm optimization-based image segmentation algorithm is proposed to improve image processing efficiency and enhance real-time visual processing.For the high specificity requirement of the system,the contour tracking algorithm is used to obtain the biological object and actuator position information within the microscopic field of view.To achieve accurate and nondestructive height positioning of the actuator,an actuator height estimation method based on the sharpness evaluation function is proposed to achieve contactless and rapid positioning of the actuator height by combining the depth-of-field parameters of the imaging device.To improve the image processing speed and reduce the leakage rate in the biological capture process,an image pyramid-based template matching algorithm is used to extract the region of interest and ignore the irrelevant information in the field of view.For the interference such as actuator occlusion and actuator surface dirt in the capture process,the image filtering algorithm is designed to achieve real-time prediction and correction of biological position and improve the detection accuracy.Second,the control strategy of the capture process is investigated.The target search and correction strategy is designed to rapidly search and localization of biological targets under the microscopic field of view.To address the problem of a complex environment and low efficiency of actuator positioning under the microscopic field of view,the path planning algorithm is optimized according to the motion environment to generate a smooth and collision-free trajectory to improve the accuracy and efficiency of control.To address the problem that the existing system neglects the damage to the target during the capture process,the theory of robot compliance control in the macro field is introduced into the micro-operation field.A force simulation method for biological objects is proposed to realize the position tracking and force adjustment of biological targets,to improve the dynamic interaction capability of the system from the control perspective,and to reduce the damage suffered by the biological target capture process.Finally,a platform is built to conduct several experiments for the proposed target localization and control strategies.Experimental analysis is conducted in terms of image segmentation,target detection,target position correction,actuator end trajectory tracking,and biological target capture.The results show that the used image processing algorithm can achieve fast and accurate visual detection,the used position correction and trajectory tracking algorithm can meet the system positioning accuracy and speed requirements,and the used capture control strategy can meet the accurate positioning while dynamically adjusting the force on the biological target during the capture process to reduce the damage to the target and improve the survival rate of the biological object. |
PDF Full Text Request