This work presents a new approach on simultaneously estimating position, strength, and spread of multiple thermal updrafts. A particle filter is proposed to estimate those as parameters of a nonlinear thermal updraft model. Given the nonparametric nature of the particle filter, the approximation of the posterior is not restricted by the assumption of a Gaussian probability distribution. In particular, the estimate can become multimodal. It is therefore possible to estimate the characteristics of multiple thermal updrafts within a single filter. In order to evaluate the resultant particle distribution, a cluster analysis suited for the application at hand is presented. Observability is analyzed by evaluating the Fisher information for the specific filtering problem. General feasibility of the filtering approach is demonstrated by presenting simulation results for a fixed-wing aircraft traveling a region with several stochastic, transient thermal cells.
展开▼
