This work is about solving the global localization issue for mobile robots operating in large and cooperative environments. It tackles the problem of estimating the pose of a robot in the environment using real-time data either from the robot on-board sensors or/and from the sensors in the environment or/and the real-time data coming from other robots. The paper focuses on the 3-DOF localization of a mobile robot that is to say the estimation of the robot coordinates (xmr, ymr , θmr) in a 2D-environment. The interest of this method lies on the ability to easily integrate a large variety of sensors, from the roughest to the most complex one. The method takes into account the following constraints: a flexible number of measurements, generic goniometric measurements, a statistical knowledge on the measurements limited to the tolerance, and the fact the measurements are acquired both from the robot onboard sensors and the environment sensors. The approach is able to integrate an heterogeneous set of measurements not only generic goniometric measurements but also range, position given by a tactile tile, complex shape, dead reckoning measurements. The way that outliers and environment model inaccuracies can be taken into account is described. The problem of nonlinear bounded-error estimation is viewed as a set inversion. The paper presents the theoretical formulation of the localization method in a bounded-error context and the parameter estimation based on interval analysis. Simulation results as well as real experiments show the interest of the method in a cooperative environment context.