In light of the complexity and nonlinearity of the end motion of the manipulator, an improved interactive multimodel algorithm based on particle filter was proposed. By comparing and analyzing the extended Kalman filter and particle filter, it is concluded that the particle filter has good filtering performance when dealing with nonlinear problems.Particle swarm optimization (PSO) algorithm was used to solve the problem of low coincidence degree between prior distribution and likelihood function distribution in the particle filter.The improved particle filter was used as a filter, outliers in the interactive multimodel algorithm that had great influence on the filtering effect were analyzed and they were improved by using adaptive robust theory.The experimental results show that by using the improved algorithm, the influence of noise is reduced and the tracking and positioning accuracy of the end of the manipulator is improved.