In multi-sensor environment when availability of measurement is uncertain, it is difficult to construct uniform global observation vector and observation matrix appropriately. To resolve the problem, a measurement fusion algorithm for uncertain measurement was presented. By defining availability function for each dimension of observation vectors to construct generalized observation vectors and covariance matrixes, the uncertainty of measurement was expressed, and formal valid measurement was obtained. The existing measurement fusion algorithm of parallel filtering was then generalized to uncertain scenario, and optimal fusion result can be obtained. To be convenient for numerical calculation, a suboptimal algorithm was put forward also. Simulation results showed that the method presented can deal with the multi-sensor measurement fusion of uncertain availability correctly and calculational cost is almost as the same as one of existing algorithm for certain measurement.
