Abstract :
One of the most successful techniques to arise from modern control theory works in the following way. Using current measurements of states of the system, control values are determined that make the output of the system follow a desired trajectory over a specified interval of time into the future. Assuming sufficient computational power and a sufficiently accurate model of the system, numerical algorithms are used to compute these control values. The computations are usually generated by an optimization method and can take into account constraints that must be satisfied, for example, on the states and controls. Of the computed control values, only the first value is implemented. The procedure is repeated at the next time instant and so forth. This method is known as either receding horizon control or model predictive control.
