API Reference

quakfilter v0.1.0 — 24 classes, 3 functions

quakfilter provides modular state estimation building blocks — Unscented Kalman Filters, Interacting Multiple Model estimators, and a multi-object track pool — all backed by a high-performance C++ core with full Python bindings.

Install

pip install quakfilter

API Sections

Filters

UKF and IMM estimators for non-linear state estimation

2 classes

Motion Models

Rigid body dynamics — constant velocity, acceleration, turn-rate, Singer, helical

8 classes

Sensors

Measurement models for pose, velocity, and projective bounding box sensors

7 classes

Tracking

Multi-object track pool with birth, death, and data association

3 classes

Utilities

Manifolds, result types, and time conversions

4 classes, 3 functions

Quick Start

import numpy as np
import quak

# Constant-velocity rigid body model
model = quak.RigidBodyCVModel()

# 16D state: position(3) + velocity(3) + quaternion(4) + angular velocity(3×2)
x0 = np.zeros(16)
x0[9] = 1.0  # quaternion w-component

ukf = quak.UKF(
    model=model,
    process_noise=np.eye(15) * 0.01,
    measurement_noise=np.eye(6) * 0.001,
    initial_covariance=np.eye(15) * 0.1,
    initial_state=x0,
)

# Predict-update loop
ukf.predict(dt=0.1)
state = ukf.update(measurement)