Research On Nonlinear Control Of Dual Hemisphere Capsule Robot

Compared with traditional endoscopes,the capsule robot has great potential in safe,reliable,and painless technology.Capsule robots can complete high-efficiency,non-invasive traversal diagnosis and treatment of the human gastrointestinal tract,and has a very broad clinical application prospect.For realizing the accurate posture control of the capsule robot,our research group has developed a dual hemispherical capsule robot(DHCR),which realizes the active and passive mode conversion of the capsule in the gastrointestinal tract,that is,rolling walk in the active mode and posture adjustment in passive mode,which improve the controllability of the capsule robot.Based on the research of our research group,this paper considers that the self-developed DHCR is a typical multi-variable,time-varying nonlinear system.The traditional linearization research method cannot reflect the non-local nature of the nonlinear system.At the same time,because the robot operates in the complex environment of the gastrointestinal tract,there are indeterminate factors such as centroid deviation,gastrointestinal peristalsis interference,and non-structural friction torque.The actual control model of the capsule robot can not be accurately established,so that it is difficult to achieve real-time precise control of the DHCR.Aiming at the above two control difficulties,this paper proposes a nonlinear control strategy of DHCR based on radial basis function neural network,which not only avoids the limitations caused by linearization,but also can estimate and compensate for errors caused by indeterminate factors of system,thereby improving the control performance of the DHCR.Firstly,the modal conversion mechanism of the DHCR is explored in combination with the actual working environment,and the posture characteristics and the orientation information of the DHCR are analyzed in depth during the fixed-point observation process.And the dynamics model was deduced,which laid a foundation for the research of the motion control strategy of the capsule robot.Then,the radial basis function(RBF)neural network compensator is introduced to approximate indeterminate factors such as unstructured environment friction,centroid offset and unknown model parameters in the system,etc.,and compensation is carried out at the control input to suppress the influence of indeterminate factors.At the same time,the robust performance of the sliding mode controller is used to overcome the approximation error of the compensator,so as to improve the control performance of DHCR.Finally,the stability of the nonlinear controller of the capsule robot is verified by the Lyapunov equation,and the designed controller is numerically simulated and the pig large intestine experiment is carried out.The theory and experimental results show that the control method based on RBF neural network compensator can effectively suppress the influence of indeterminate factors on control error and improve the control accuracy of DHCR,and the dynamic performance of the DHCR in an unstructured environment is improved,which lays the foundation for the accurate diagnosis and treatment of the gastrointestinal diseased area.