Research On Control Methods Of Flexible Needle System Based On Online Estimation

Punctures have important implications in medical treatment, diagnostics and scientific research. It is also considered the easiest and most minimally invasive medical procedures. Robot-assisted puncture system, which belongs to the minimally invasive surgical robot, receives wide attention from researchers and doctors because of its accurate positioning and radiation protection. However, the current puncture needle is steel needle, the path of which is straight and cannot avoid blood vessels, nerves and other sensitive tissue. To solve those problems in the application of steel needle, this thesis carried out a research on robot-assisted flexible needle puncture. The appropriate control of flexible needle can not only improve the puncture accuracy, but also make the needle tip reach the target that steel needle can’t capture.Firstly, this thesis researches on planar control of flexible needle under the limitation that medical imaging equipment, such as Computed Tomography (CT), ultrasound, cannot get orientation information of needle tip directly due to small size of the tip. The nonlinear plane model is got based on nonholonomic kinematic model of flexible needle by dimensionality deduction. Then, use the direct feedback linearization to control the flexible needle. At the same time, the orientation of needle tip is estimated online by luenberger observer. Then, feedback control law is designed based on estimated states. To enhance the robustness of the system to noise, Linear Quadratic Gaussian (LQG) based planar control is proposed. This method estimates states online using Kalman filter, and designs the optimal feedback control law by linear quadratic regulator (LQR).Secondly, in order to overcome the weakness of feedback linearization that noise has great influence to control law, the plane model of flexible needle is linearized locally. Based on the linearized model, this thesis presents two other planar control methods using extended kalman filter (EKF) or unscented kalman filter (UKF) respectively, which are nonlinear filtering methods based on statistical properties of the noise. The result of simulation shows that this method can improve estimation accuracy of the system states and the control performance as well.Thirdly, by combining planar control and two-dimensional path planning, this thesis presents three-dimensional control strategy based on plane switching. The position and orientation of flexible needle related to two coordinate systems, which are fixed on two planes, are analyzed seriously, and the initial position and orientation are adjusted to zero. Then, this thesis conduct path planning in two planes respectively.In the end, experiment platform of the robot-assisted flexible needle puncture is introduced and system noise produced by binocular camera is analyzed. Theoretical methods are verified through experiments.To sum up, the control of robot-assisted flexible needle puncture are carried out comprehensively. By combining planar control and two-dimensional path planning, a three-dimensional control strategy of flexible needle is proposed. Experiments achieved good results which further pave a solid pathway for clinical application with robot assisted flexible needle puncture technology.