In this project, I implemented an Extended Kalman Filter (EKF) to perform 2D localization of a mobile robot navigating a forest of cylindrical obstacles. The robot’s pose was estimated by fusing wheel odometry (translational and rotational velocity) with range and bearing data from a front-mounted laser rangefinder, both affected by realistic Gaussian noise. Using a nonlinear motion model and a landmark-based observation model, the EKF incorporated time-varying landmark visibility, measurement gating, and variable observation counts per timestep. Sensor variances were derived from ground-truth Vicon motion capture data. The implementation was tested under multiple sensor visibility thresholds and initial conditions to evaluate estimator convergence and uncertainty behavior in a noisy, partially observable environment.