GPS/IMU Sensor Fusion 2D: Vehicle Localization Drift

This is the 2D top-down counterpart to the 3D GPS/IMU sensor-fusion simulator, running the exact same predict/update Kalman math on a flat canvas so the numbers are easy to read at a glance. A vehicle drives a closed loop while three independent position estimates race each other: a raw IMU dead-reckoning trail that integrates noisy acceleration and drifts steadily off course, a GPS-only trail that holds the last noisy fix until the next one arrives, and a Kalman-fused estimate that combines both to track the true path far more closely than either sensor alone. A live error-vs-time chart plots all three continuously, and a simulated tunnel cuts GPS entirely so you can watch the fused estimate's uncertainty inflate during the blackout and collapse the instant a fix returns.