An optical trap uses radiation pressure of light to manipulate microscopic objects. The interaction between the light and the microscopic objects result in the objects experiencing optical forces. These forces are on the same order of magnitude as biological forces (typically SIrange[range-units=single]{0.1}{100}{picoewton}) and this feature makes optical traps appropriate for single-molecule studies. Currently, there is a growing need to create an automated optical trap that uses the entire operating range of the optical trap to study the biological forces. Spatial nonlinearities in the optical force and parameter uncertainty complicate feedback control for optical traps. A consequence is that users are spending an enormous amount of time calibrating the instrument and designing a controller, and this diverts their time away from studying the biophysics. This research explores the use of nonlinear and adaptive feedback methods to create an automated optical trap. ududA model is defined to describe the coupling between the dynamics of the optical trap and molecule, and the nominal force within the molecule is treated as a disturbance. The disturbance information is obtained by creating a disturbance model and combining its dynamics with the system dynamics. The system nonlinearities are addressed by using a nonlinear Kalman filter to estimate the system state, then the system state is used in a input-output feedback linearization and linear quadratic structure to satisfy performacne requirements. Statistical analyses are performed to assess the effectiveness the feedback methods have on the open-loop and closed-loop systems. Its performance is compared with that of linear integral control used in practice to quantify the performance improvement when considering the system nonlinearities in the control design. The system nonlinearities and parameter uncertainty are addressed by using adaptive and nonlinear feedback methods. An adaptive state observer provides a simultaneous estimate of the system state and parameters, then these estimated entities are used in an adaptive input-output feedback linearization and LQ structure. The result is the creation of an automated self-tuning optical trap that minimizes the user interaction with the instrument calibration and control design, uses the entire operating range of the optical trap, and obtains an unbiased estimate of the molecule force. The closed-loop performance of these feedback methods are demonstrated by replicating the force-extension curve of a DNA molecule.
展开▼
