Performance of the multi-objective particle swarm optimization (MOPSO) is improved by devising a novel archiving strategy. Premature convergence by Monte Carlo localization during the localization process can be easily detected by three proposed rules. Premature convergence problem in global localization for mobile robots in highly symmetrical environments is overcome by the proposed approach. Global localization performances in terms of success rate and computational time in highly symmetrical environments are significantly improved by the proposed approach.