In practical indoor environments, wheels experience slippage and wear, which leads to cumulative errors in wheel-based odometry and, consequently, a decrease in global positioning and pose tracking accuracy. To address this issue, an improved method has been proposed that enhances the localization accuracy of mobile robots by using 2D laser scan points to match with a grid map. This method first uses the pose estimated by the Adaptive Monte Carlo Localization (AMCL) algorithm as the initial pose and then matches the laser scan points with the pre-constructed grid map. The pose increment is obtained by solving it with the Gauss-Newton method, resulting in more accurate localization. Experimental results validate that this algorithm effectively corrects the pose and improves localization accuracy.