Abstract: Autonomous navigation system based on ultraviolet sensors is a typical nonlinear system.For the general particle filter lacks the adaptive capacity.A new particle filtering algorithm called adaptive particle filtering was proposed which adopts a new method combining the unscented Kalman filter with the strong tracking filter to produce important density functions.The proposed algorithm adopts UKF to produce particles,in which each sigma point of each particle was updated by STF to make the algorithm have adaptive.Simulation was done based on simulated orbit and measurement data and was compared with results of other filtering algorithms to illuminate the effectiveness of the navigation method.