Research On The Manipulation Of Softtissues Based On Visual Servo

The percutaneous puncture is an important method of the minimally invasive surgical. Instead of the traditional stiff short needle which is hard to meet the needs for the obstacle avoidance in the puncture process, a long and flexible needle with a bevel tip has come out and becomes a research topic, of which the shape of needle trajectory could be a curve. The trajectory of the flexible needle is very flexible and multiple. However, a flexible needle insertion with a big curve trajectory would probably tear the tissue for the elastic deformation of the needle, while one with a linear trajectory may bruise the tissue because of the high rotation speed of the bevel tip. To solve this problem, the "needle-tissue" compound operation technique is adopted, which can shift the target point to simplify the insertion trajectory and reduce the damage by controlling the deformation of the target tissue.However, most of the methods of the active deformation control require a priori knowledge of the object’s deformation properties. The deformation parameters of human tissues in the application of needle insertion are unknown, for their individual differences. In this thesis, the uncalibrated visual servo technique is adopted. By estimating the unknown deformation Jacobian matrix which containing the unknown deformation parameters and the camera projection model, the system can compute the estimated deformation Jacobian matrix, with which the velocity controller can determine the output control velocity. Thus, the deformation feature’s difference computed by the estimated deformation feature predicted by the estimated deformation Jacobian matrix and the true deformation feature measured by the camera can be used to update the estimator. At last, the stability of the control algorithm is proved by the Lyapunov’s stability theorem, and the performance indexes of the control algorithm are analyzed through the simulation experiment.This thesis puts forward the tissue’s deformation control for the flexible needle insertion. Before the deformation control, we need to do the deformation control planning to obtain the target shift. Since the target shift is used to simplify the trajectory, it can be determined by the insertion trajectory planning. However, the insertion trajectory can be composed of various types. In order to solve the target shift planning problem, the plane constant curvature arc which shaping in homogeneous and isotropic object tissues only is adopted to design the trajectory in this thesis. The solution of the insertion trajectory with the constraint condition of multiple obstacles is deduced, and the cost function is put forward to assess the insertion trajectory. Considering that it is a multi-objective optimization problem, the improved NSGA2 genetic algorithm is used to solve it, and the cost function is modified by the optimum hyper-plane of the final population. Finally, the multi-objective optimization can be transferred into the single objective cost function by the weight coefficients to obtain the optimum insertion trajectory, which can be used to determine the deformation control plan.Finally, the software system of the deformation control experiment is compiled, including the target detection module, the robot control module, the control algorithm and so on. In this thesis, the existing image processing methods are compared and analyzed, and the target detecting and tracking is achieved by the Open CV. The whole process of the deformation control can be divided into two parts: one part is to obtain the target shift plan through the trajectory optimization according to the specific environment, and the other part is to manipulate the object tissue to the predetermined deformation target through the deformation control system. In the end, the experimental data is analyzed to verify the validity of the control theory and the feasibility of the deformation control system.