Local path replanning for UCAV based on mix-state dynamic Bayesian networks

The principle to construct local path diagram with sudden appear threats is established, and a local path replanning scheme for unmanned combat air vehicle(UCAV) is developed. Firstly, constructing and searching an improved voronoi diagram based on the locations of the different threats, an initial reference threat-avoiding flight path to the target is generated using Dijkstra algorithm. Secondly, switching linear dynamic system(SLDS) based on Mix-state dynamic Bayesian networks(DBN) is used for apperceiving battlefield's situation change. The location and threat of sudden appear threats are estimated using Viterbi approximation algorithm. The next, a best local path can be found out using base principle. In the end, cubic spline theory and sequential quadratic programming are used for optimizing the initial reference path. The optimized path is flyable to the unmanned combat air vehicle. The Matlab simulation result demonstrates the local path planning algorithm is effective.