×

Mobile robot Monte Carlo localization based on improved unscented particle filters. (Chinese. English summary) Zbl 1240.68411

Summary: Particle filter is a key issue in mobile robot Monte Carlo location (MCL). Firstly, an improved unscented particle filter (IUPF) algorithm is proposed in this paper. To overcome particles degeneracy, the algorithm utilizes an iterated sigma points Kalman filter to generate a more accurate proposal distribution, which introduces most recent measurement information into a sequential importance sampling routine through iterated update processing. Secondly, by applying IUPF to MCL, an IUPF-MCL algorithm is given. Finally, simulation results show that IUPF-MCL is an accurate and robust mobile robot localization algorithm.

MSC:

68T40 Artificial intelligence for robotics
93E11 Filtering in stochastic control theory
60G35 Signal detection and filtering (aspects of stochastic processes)
PDFBibTeX XMLCite
Full Text: DOI