Non-model based estimation with applications to sensor calibration using multiple sensors

A nonmodel based estimation procedure is introduced. The best estimate is defined to be the geometric center of the intersected error volume, and the error of the best estimate is defined as the distance of the intersected error volume measured from the center. Khachian's algorithm is extended to find the best estimate. When the algorithm is viewed as a system of nonlinear difference equations, conditions are established to test the existence of periodic solutions. A stopping rule is also introduced. A strategy for finding the error of the best estimate is described. Examples are given to illustrate the estimation procedure.