No abstract

This paper proposes a novel solution to the fixed-ground target tracking control problem of satellites utilizing a Nonlinear Model Predictive Control approach (NMPC). The Continuation / Generalized Minimal Residual (C/GMRES) algorithm is selected as a promising fast solver to an optimal control problem in real time. The algorithm could perfectly deal with the huge computational load of this approach, represented in solving Riccati differential equation, by simple and efficient approximations. A new controloriented model converting the main tracking problem into a simple regulation problem is developed. This simple and easy traceable reformulated model has an advantage in dealing with modeling errors and unplanned external environmental disturbances. The update of the control input is obtained by integrating a deduced time-dependent inputs and Lagrange multipliers vector; representing the solution of a set of linear equations and corresponding to the optimality conditions. The proposed algorithm is simulated using real satellite parameters to track a fixed-ground target for reconnaissance purposes. The simulation results show that the algorithm of C/GMRES method can track a desired fixed ground target robustly, with precise tracking error and guaranteed safe stability limits for shooting activities throughout the overpass flight.

No abstract

Although solving inverse dynamics problems is performed a lot in literature, none of the previous references addressed the general unrestricted solution of three dimensional (3D) trajectories of guided gliders. This glider could be any subsonic flying body such as guided ammunitions. Inverse dynamics could be one of the key techniques of solving similar problems. In this paper, 3D trajectory generation and following are performed. The trajectory generation is divided into three phases: heading correction, glide and terminal phases. The solution of the inverse problem is performed analytically, using the Mu-Pad tool. Next, a special formulation of the general dynamics equations enables the solution of such a problem and the calculation of the required deflection angles. Finally, a six degrees of freedom (6DoF) direct simulation is performed using the obtained deflection angles in order to compare its trajectory with the generated one. This comparison yields fairly good results and validates the quality of the proposed inverse dynamics solution.

Parallel Robot (PR) has shown its ability to be precise in its movement. Actuators move simultaneously to achieve the required target, on top of that its payload is much greater than what a serial robot can withstand. To determine workspace of the robot with known angles Forward kinematics has to be introduced which, bring a lot of difficulty as it requires the solution of multiple coupled nonlinear algebraic equations. Those equations bring multiple valid solutions. Those solutions could lead to different locations. As it is not going to make the pick and place for PR will be easier. This paper will discuss a numerical method that calculates the Forward Kinematics for PR. This method uses Artificial Neural Network which relay on training with a certain number of iterations. The set of data to be used in the training can be obtained from PR simulation. This method will serve to know workspace around PR as it will help it to pick the target object.