Iterative Learning Embedded Composite Model Reference Adaptive Control for Off-Axis In-Situ Rotation in Nanorobotic Manipulation

Realizing dexterous rotation under the microscope has been a great and yet not-well-addressed challenge in micro/nano-robotics. The limited field of view (FOV) of a microscope necessitates point-wise in-situ rotation, which refers to manipulating objects within their existing or natural environment. Though on-axis in-situ rotation has been investigated, off-axis in-situ rotation, which is more suitable to classical nano-robot setups, remains unexplored. The main difficulty facing here is the model uncertainties and visual disturbances that fail traditional control methods. In this letter, we propose a novel control scheme, namely the iterative learning embedded composite model reference adaptive control (IL-CMRAC), for efficient visual servo to solve the problem of in-situ off-axis rotation, by fully leveraging the repetitiveness nature of rotational motion. IL-CMRAC takes advantage of both offline learning and online adaptation to tackle the difficulties brought by uncertainties. Experiments demonstrated that the proposed method is capable of realizing high-precision off-axis in-situ rotation by average error within several microns.