The role of total least squares in the common attitude estimation problem

Single point attitude determination is the problem of estimating the instantaneous attitude of a rigid body from a collection of vector observations taken at a single moment in time. Traditionally, vector measurements are compared to some known model that is assumed to be error free. Several solutions have been developed since Wahba first proposed the problem. Each technique takes a unique approach to the attitude estimation problem and solutions can be expressed using a variety of attitude parameters. More recently, the relative attitude estimation problem has gained attention which assumes measurement errors are present in both sets of vector observations. The discussed techniques use a familiar least squares approach, but the nature of the relative attitude problem suggests that a total least squares approach is more appropriate. Thus, the role of total least squares in relative attitude estimation is investigated through numerical examples.

The role of total least squares in the common attitude estimation problem Conference Paper