A Fast Conjunctive Resampling Particle Filter (FCR-PF) for the Multi-Robot Localization problem is proposed. The idea is to exploit a plain Particle Filter to keep low the computational load and focus the attention to the collaboration among robots in order to improve the estimation capabilities. Collaboration can be setup at any time when two robots are within their range of visibility. It consists of exchanging sensory data along with the most significant subset of the particle distribution. These information are exploited both locally by each robot to solve environmental ambiguities and globally to refine the estimation process. In this way, a broader exploration of the state space is achieved with a significant increase of the convergence rate, while the asymptotic computational complexity is kept the same as for the plain Particle Filter. Simulations have been executed to underscore the effectiveness of the proposed collaborative localization algorithm.