Identification and design of experiments in a quasilinear regression problem

The paper is concerned with evaluation of the parameters of a quasilinear regression model in controlled experiment, the data on the mathematical description of the experiment possibly incomplete. The results are a recurrent estimate of the parameters which optimizes a minimax criterion, a recurrently computed upperbound of the covariance matrix, and algorithms for choice of the design of the next experiment for optimizing the evaluation performance on the knowledge of current observations.