The robot localization is a crucial task that needs to be solved as a part of the navigation problem for an autonomous robot. In order to estimate the location of a robot in the environment, various sensors are used to extract meaningful information from measurements to acquire knowledge about the robot’s environment and motion. Due to the fact that the sensor uncertainty is random, it is impossible to find an accurate pose for the robot by only one senso,r and the accuracy of any sensor is generally related to its price. Sensor fusion technique is a well•known approach to give the best estimate robot location and how certain it is by combining data from two or more inexpensive sensors. In this paper, the estimation of the robot’s new pose given the previous pose and error-accumulated odometry is proposed based on the fusion of data from wheel encoders and the Inertial Measurement Unit (IMU) for a differential drive mobile robot. The robot has a very simple driving mechanism that is quite often used in practice, especially for service mobile robots. The required mathematical models of the robot and indoor localization system are derived. The mathematical tool for sensor fusion is the Kalman filter which provides the optimal estimate of the system state, and robot configuration, assuming that the noise from each sensor is zero•mean and Gaussian. The robot was driven in two different cases; circular trajectory and square trajectory to evaluate the performance and consistency of the robot localization. The experimental result shows the effectiveness of the proposed work for the robot’s pose estimation.