Abstract: In order to use the autonomous mobile robot to search for unknown radioactive objects, a method for locating the radioactive source with improved particle filtering was proposed based on the recursive Bayesian estimation model. Firstly, an initial particle set was established, and the weight of the particle was updated and normalized according to the observation value. Secondly, the auto-optimal resampling method was introduced to increase the particle diversity in the resampling process. Finally, the particle carrying out the convergence condition was weighted and summed to estimate the position and activity parameters of the radioactive source. The simulation results show that the method is feasible and effective. The positioning accuracy is high in the unshielded environment, and the approximate position of the radioactive source can also be found in the shielded environment, which provides a reference for the final location of the radioactive source.