Mechanism for preventing premature convergence in robot localization

In this paper, we propose a new approach to solve the premature convergence problem in Monte Carlo Localization for global localization, entitled "a mechanism for preventing premature convergence". The algorithm uses a referenced relative vector to rearrange weight for each sample, allowing better exploration of a symmetrical environment, resulting in preventing premature convergence problem to converge particles successfully. Simulations have confirmed that the proposed approach can offer notable improvement in the localization performance.