Aiming at problem of low tracking precision of particle filtering（ PF） algorithm caused by quantity and quality effect of particle

and the importance density function can not be obtained directly and noise can not be optimized in resampling process

which give rise to particle degeneration

propose a genetic-variance adaptive PF（ GVAPF） algorithm. Firstly

use genetic algorithm to select initial particles from large numbers of particles in order to improve quality of initial particles. And then

modify noise in resampling process by variance adaptive（ VA） technology in real-time

and make the importance density function gets closer to true distribution of the state. Simulation results show that the improved algorithm is superior to the standard PF algorithm.