Command-Line Interface

The kalbee CLI is a zero-code way to see the library work — no Python script required. Core commands (demo, bench) need nothing beyond kalbee's normal numpy/scipy dependencies; the animated --live view needs rich: pip install kalbee[cli].

kalbee demo

Runs a filter against a synthetic signal and reports tracking accuracy.

kalbee demo --filter akf --signal maneuver --steps 200

Add --live for an animated terminal chart (true signal, noisy measurement, and filtered estimate, each as a live-updating sparkline):

kalbee demo --live --filter kf --signal sine

Flag	Default	Meaning
--filter	kf	One of kf, akf, srkf, if, vkf, fmkf, hinf (filters that share the plain (F, Q, H, R) constructor this demo builds).
--signal	sine	sine, cosine, linear, step, or maneuver.
--steps	150	Number of time steps.
--noise	0.3	Measurement noise std.
--process-var	0.1	Process-noise variance.
--live	off	Animate in the terminal (needs kalbee[cli]).
--fps	15	Frames per second in --live mode.

kalbee bench

Runs the built-in filter benchmark suite (speed + accuracy across all core filters) and prints a sorted comparison table.

kalbee bench --duration 10 --dt 0.05

For a comparison against FilterPy/pykalman/simdkalman on identical data, see Benchmarks and scripts/compare_benchmarks.py.

kalbee new

Scaffolds a runnable starter script so you don't start from a blank file.

kalbee new my_tracker.py                       # basic 1-D KF tracking loop
kalbee new gps_fusion.py --template gps_imu     # GPS + IMU recipe
kalbee new attitude.py --template attitude      # quaternion attitude EKF recipe

Pass --force to overwrite an existing file.