Constrained state estimation for underwater target tracking problem

To enhance estimation accuracy, range, and velocity limits are used to perform constrained state estimation. The range limits are determined using machine learning techniques with the help of bearing angle, Doppler-shifted frequency, and intensity of acoustic signals received at the observer sonar as inputs. The Doppler-shifted frequency from the target can be used to determine its velocity limits. These range and velocity upper and lower limits are used as constraints while performing state estimation. The optimization problem is solved using the Lagrange multiplier. The proposed method is implemented on a bearings-only tracking problem and a Doppler-bearing tracking problem using a moderately nonlinear and a highly nonlinear scenario. The proposed estimation method is observed to have more estimation accuracy than the state-of-the-art and traditional filters in terms of root mean square error, average normalized estimation error squared, bias norm, track loss percentage, and relative execution time.