Research On Position Tracking Control Of Non-linear Robotic Arms

The main focus of this thesis is to develop a control strategy to improve the tracking performance of nonlinear robotic arm as compared to existing control strategies proposed in the literature.Since the robotic arm exhibits two types of motions;the one is repetitive motion and other is non-repetitive motion.So this research is focused on the design of separate controllers to deal with the two types of motions of the robotic arms.Overall,the thesis is divided into two parts:the first part proposed a fractional order control strategy to deal with non-repetitive motion and the second part presented iterative learning control strategy that is capable to deal with the repetitive motion of non-linear robotic arm.(1)In the first part,the fractional order PID controller is designed to achieve precise position tracking of nonlinear robotic arm.Since the dynamics of real-time system is highly nonlinear and there exists a coupling between the states that increase the complexity of overall system so it becomes a complex task to design an appropriate controller directly on the nonlinear system.In this report,fractional order modeling technique is adopted that is considered more accurate representation of nonlinear system dynamics as compared to linear time-invariant and linear time-varying model of systems.The global approach of model identification is followed here in which input-output data from the nonlinear system is used as input to the model identification algorithm.The gradient-based Marquadt algorithm is used to identify the unknown parameters.Two fractional order models are identified that represent the dynamics of both links of the robotic arm.The identified models are validated on nonlinear model of the system.Magniton stability criteria is adopted to check the stability of fractional order system.(2)The next task is to design fractional order PID controllers that are based on the identified fractional order model of system.FOPID controller has many advantages over conventional PID controller such as increase in stability margin,steady state error elimination,robust behavior against the parameter variations and high-frequency disturbance and output disturbance elimination.In the design of fractional order controller,optimal tuning of controller parameters is another milestone to deal with such that desired performance can be achieved.However,in this work,Nelder Mead optimization algorithm is adopted to tune the controller parameters because of fast computation and ability to deal with high dimensional problems as compared to particle swarm optimization.The performance criteria is formulated here based on time domain performance specifications using weighted sum approach.The designed controller is then implemented on the nonlinear model of robotic arm.The simulation results are compared with existing PID controller design on the same system.(3)Similarly,a new control strategy proposed for the position tracking of robotic arm is the modified computed torque controller.The nonlinear system dynamics is linearized using inverse dynamics of the model and fractional order PID controller is designed to deal with remaining modeling errors.The optimal values of controller parameters are calculated using Nelder-Mead optimization technique based on desired design criteria.The control design consists of two controllers,one in the inner loop i.e.,computed torque controller and one in the outer loop i.e.,FOPID controller.The improvement in the tracking performance as compared to conventional PID controller in the outer loop is shown by the simulations.(4)The second part of the thesis is based on iterative control strategy that has been used extensively in repetitive motion control of robotic arms.A slight modification is done in the existing point to point learning approach in which the problem of varying initial conditions is solved by neural network state learning method.Whereas,the receding horizon control scheme is combined with the point to point ILC to improve the tracking performance of point to point ILC.The results are supported by the simulations.