Autonomous Manipulation Of Electromagnetic Driven Microrobots

Micro-robots wirelessly controlled by the low-intensity electromagnetic field have a small size and can move in the confined space to complete complex operations towards biomedical applications.Various researches on electromagnetic manipulation of micro-robots have been implemented over the past several decades.However,the current manipulation techniques are mostly designed under ideal conditions,without considering the disturbances from the gravity and boundary,and lacking of effective motion planning solutions.Therefore,this thesis aims to establish an autonomous manipulation solution to plan motion of micro-robots effectively and perform sub-millimeter precision control.Firstly,a self-made three-axis Helmholtz electromagnetic coil with vision system is designed to track the helical micro-robot and implement path following task.Helical micro-robots are made of flexible helical plant fibers which are covered by magnetic Fe₃O₄ nanoparticles.Theoretically,the swimming model of the helical micro-robot with the bending body is established.The bending deformation characteristics of the helical micro-robots during the swimming are also studied experimentally.It is found that the bending body contributes to improve the swimming speed.Secondly,a motion planning scheme and a stable motion control method in a complex planar environment are developed.The sampling-based motion planning method is applied to the micro-environment to calculate the feasible motion path.Moreover,the search efficiency among different motion planning methods is also discussed.The extreme learning network is employed to estimate the relationship between the motion of the helical micro-robot and the driven magnetic field.A stable motion controller is developed based on the sliding-mode variable structure combined with PI control.In order to solve the problem of controller saturation caused by the jump signal,a bioinspired filter model is introduced to smooth it.Thirdly,an autonomous manipulation system including motion planning scheme and motion control method in three-dimensional?3D?space is developed.The optimized sampling-based motion planning method is presented to calculate the feasible path towards solving the issues of heavily computing in high-dimensional space.An off-line learning model and an online learning model are formulated to approximate the motion of helical micro-robots in 3D space.The off-line learning model uses error backpropagation to optimize the parameters of neural network.Furthermore,the online model can update the model parameters to solve the problem of large estimation errors using the off-line method.The decoupled motion controllers are developed and their stability are also analyzed based on the Lyapunov theory.In summary,using the self-made electromagnetic driven platform,the autonomous manipulation scheme of the helical micro-robot is completed with the sub-millimeter manipulation precision,this lays the solid foundation for the motion planning and precise control of the targeted movement towards biomedical applications.