Active Tremor Compensation in Handheld Instrument for Microsurgery

Abstract

Human?s ability to perform precise micromanipulation is limited by small involuntary movements inherent to normal hand motion. Microsurgery is one area where the surgeons? performance is hampered by this manual imprecision. Not only that it complicates many delicate surgical procedures, it also makes certain types of intervention impossible. The most familiar type of erroneous movement affecting a healthy person is physiological tremor. Instead of going with the more familiar approach of using a teleoperated robotic system, we adopted a less obtrusive and much cheaper approach of implementing accuracy enhancement within a completely handheld tool. This device senses its own motion, distinguishes the erroneous motion from the intended motion, and manipulates its own tip in real time to compensate the erroneous motion. This dissertation focuses on the sensing and compensation of the erroneous motion, while tremor modeling and estimation is performed by a previously developed weighted-frequency Fourier linear combiner (WFLC) algorithm. Instantaneous motion of the instrument is sensed by a new magnetometer-aided all-accelerometer inertial measurement unit (IMU). The sensing system consists of three dual-axis miniature accelerometers and a three-axis magnetometer. The redundancy in sensing provides two sources of orientation and position information. The angular motion information derived from the differential sensing kinematics algorithm has very high sensing resolution but suffers from integration drift; on the other hand, the orientation obtained from the gravity and magnetic North vectors is noisy but non-drifting. These two complementary sensing sources are fused via an augmented state quaternion-based Kalman filter to yield high quality sensing. The instrument tip is manipulated by a three DOF parallel manipulator driven by piezoelectric actuators. The hysteretic non-linearity of the piezoelectric actuator is modeled and linearized by a rate-dependent Prandtl-Ishlinskii operator. Based on the dynamic hysteresis model, an open-loop inverse feedfoward controller is implemented to accurately track dynamic motion profiles. While the targeted application of this dissertation is in microsurgery, the principles of the approach is universal and can be extended to other micromanipulation tasks, such as cell manipulation in the biotech industry, gun-sights or handheld military tracking equipment, and handheld video photography etc.