Global Vision Localization of Indoor Service Robot Based on Improved Iterative Extended Kalman Particle Filter Algorithm

Localization is the primary problem of mobile robot navigation. Monte Carlo localization based on particle filter has better accuracy and is easier to implement, but there is also the problem of particle degradation. In this paper, the iterative extended Kalman filter is optimized by the Levenberg-Marquardt optimization method. An improved particle filter algorithm based on the upon optimized iterative Kalman filter is proposed, and the importance probability density function of the particle filter is generated by the maximum posterior probability estimation of the improved iterative Kalman filter. Simulation results of the improved particle filter algorithm show that the algorithm can approximate the state posterior probability distribution more closely with fewer sampled particles under the premise of ensuring sufficient state estimation accuracy. Meanwhile, the computation is reduced and the real-time performance is enhanced. Finally, the algorithm is validated on the indoor mobile service robot. The experimental results show that the localization algorithm’s accuracy meets requirement for real-time localizing of the restaurant service robot.