Skip to content

README.md

Latest commit

 

History

History
130 lines (86 loc) · 3.37 KB

README.md

File metadata and controls

130 lines (86 loc) · 3.37 KB

Examples

This directory contains small executable examples that demonstrate common PyRecEst workflows.

Run examples from the repository root after installing PyRecEst or after installing a development checkout.

Basic examples

Gaussian multiplication

basic/gaussian_multiplication.py multiplies several two-dimensional Gaussian distributions and checks the result against the closed-form information representation.

Run it with:

python examples/basic/gaussian_multiplication.py

Kalman filter

basic/kalman_filter.py runs a one-dimensional constant-velocity Kalman filter. It demonstrates how to:

  • import arrays through pyrecest.backend;
  • initialize KalmanFilter with a mean and covariance;
  • call predict_linear and update_linear; and
  • read the current point estimate and covariance.

Run it from the repository root with:

python examples/basic/kalman_filter.py

Kalman filter with model objects

basic/kalman_filter_with_models.py runs the same constant-velocity Kalman filter as basic/kalman_filter.py, but defines reusable linear-Gaussian model objects and passes them to predict_model and update_model.

Run it from the repository root with:

python examples/basic/kalman_filter_with_models.py

Unscented Kalman filter with model objects

basic/ukf_with_models.py demonstrates additive-noise transition and measurement model objects with UnscentedKalmanFilter.

Run it from the repository root with:

python examples/basic/ukf_with_models.py

This example follows the current backend limitations of UnscentedKalmanFilter.

Particle filter with model objects

basic/particle_filter_with_models.py demonstrates a particle-filter loop with a sampleable transition model and a likelihood-based measurement model.

Run it from the repository root with:

python examples/basic/particle_filter_with_models.py

Multi-target tracking

basic/multi_target_tracking.py runs a small linear/Gaussian multi-Bernoulli-tracker scenario with two labeled targets, missed detections, and clutter measurements.

Run it with:

python examples/basic/multi_target_tracking.py

This example currently requires the NumPy backend.

SCGP measurement reliability

basic/scgp_measurement_reliability.py runs one full star-convex Gaussian-process tracker update with per-measurement reliability weights and an active-measurement mask. It demonstrates how to down-weight partially reliable extended-object measurements and skip unsupported measurements without changing the measurement array.

Run it with:

python examples/basic/scgp_measurement_reliability.py

von Mises-Fisher multiplication

basic/von_mises_fisher_multiplication.py multiplies two von Mises-Fisher distributions on the unit sphere and verifies the analytic product relation.

Run it with:

python examples/basic/von_mises_fisher_multiplication.py

Backend selection

Select a non-default backend by setting PYRECEST_BACKEND before running the script. For example, on a bash-compatible shell:

PYRECEST_BACKEND=pytorch python examples/basic/kalman_filter.py

Install the matching optional dependency extra before using a non-default backend. See backend compatibility for known backend-specific limitations.

Notebooks

notebooks/ is reserved for interactive examples.