ExtendedKalmanFilter5D — Interactive Demo

Truth & process

v_x,0 5.0 m/s

v_y,0 3.0 m/s

σ_a (process) 2.0 m/s²

σ_b (clock RW) 3.0 m/√s

total time T 300 s

GPS satellites · per-sat scalar pseudoranges

σ_ρ (shared) 10.0 m

Each pseudorange is its own scalar measurement — the EKF processes them one satellite at a time with a 1×5 Jacobian and a scalar gain. Toggle a satellite off to see the filter degrade gracefully (the black-box NLLS solver in Block 6 needed all four to produce anything). Drag a satellite on the sky compass to change the geometry.

t = 0.0 s

Live trajectory · 2D top-down view, ±150 m window

true trajectory EKF estimate 95% position cov ellipse flashing satellite dot in corner = update arrived

Per-state estimation error · ±95% bounds

Position estimate

—

truth: —

Velocity estimate

—

truth: —

Clock bias estimate

—

truth: —

Active satellites

—

σ_p=—