Research On Decentralized Tracking Control Of Reconfigurable Manipulator Based In Constrained Environment

Compared to conventional robotic arms,which are fixed configurations and can only work in a single environment and lack practicality and economy,reconfigurable robotic arms have attracted much attention in recent years.The study of force/position dispersion control of reconfigurable robotic arms under environmental constraints is essential.The following aspects of the control of reconfigurable robotic arms have been investigated.Firstly,the research background of this topic was introduced and the significance of the research was stated in the context of the research,then a systematic discussion of the current state of the art of the subject at home and abroad was initiated,and an introduction to various aspects of adaptive neural networks was given,followed by the arrangement of the chapters in the full text.Next,the dynamics of the robotic arm was solved using the Newton-Euler iterative method,thus establishing a model of the dynamics of the robotic arm in free space,which has been taken into account friction and perturbation.On the basis of this model,a neural network controller was designed,a robust term was added to the neural network control law in order to reduce jitter,and lyapunov's theorem was used to verify stability,and finally a simulation was carried out to verify that tracking control was achieved for the two-degree-of-freedom robotic arm.Then,based on the relationship between the contact force of the end-effector in task space and the joint moment in joint space,the mapping relationship between the end-effector force and the joint moment was added to the already established dynamics model in free space,so as to derive the dynamics model of the robotic arm under environmental constraints,the constraints were divided into static constraints and time-varying constraints,and then to simplify the controller design process,the dynamics model was in the position control,the model information was divided into a known part and an unknown part,and the unknown part was approximated by an RBF neural network,and an additional adaptive compensator was added to the controller part in order to neutralise the effects of external disturbances in the friction force and the estimation errors of the neural network.Finally,the effectiveness of the controller was verified using different configurations.Finally,a concluding note was given throughout the text,discussing areas for further improvement and research and providing future perspectives on these areas.