I am solving a least square problem:Ax=b as A is a MxN matrix where M>N. x=[Tx, Ty,..... sin(theta_1), sin(theta_2).....]
I got the estimated x using pseudo-inverse. However, the result doesn't make sense as estimated some variable among sin(theta_1), sin(theta_2)..... are greater than 1. Is there any way to set up constraint on the least square estimation like: -1
I searched for "constrained-least-square" but I don't have equality condition.
The SVD isn't helpful in solving least squares problems with additional constraints such as bounds on the variables. There are iterative algorithms for finding solutions to least squares problems with such constraints. In this case, the constraints are simple bounds on the variables, so you have what is called a "bounded variables least squares problems." There are lots of software packages for solving such problems.