Estimation using subjective knowledge with tracking applications

Within the framework of classical estimation theory, there is no technically legitimate way to utilize knowledge that cannot be codified by either deterministic or strict probability models. The authors introduce a theoretically defensible approach called coordinated objective/subjective estimation (COSE) for the simultaneous incorporation of both objective and subjective knowledge in estimation. Also discussed is a technique, called heuristically constrained estimation (HCE) which is a particular interpretation of the Bayesian use of subjective priors. COSE, HCE, and classical maximum a posteriori probability (MAP) estimation are applied to a tracking problem. >