rlabbe/filterpy
Filterpy
filterpy is a toolkit for Bayesian state estimation, Gaussian statistical analysis, and time-series noise reduction. It provides a library of linear and non-linear Kalman filters, as well as routines for non-Gaussian state estimation and signal smoothing.
The project implements a variety of estimation methods, including particle filtering using Markov Chain Monte Carlo and resampling, and discrete Bayes filtering. It also includes a suite of algorithms for refining historical state estimates through backward and fixed-lag smoothing.
Additional capabilities cover multivariate Gaussian analysis using Mahalanobis distance and covariance ellipses, as well as system modeling utilities for generating noise matrices and discretizing differential equations.
Features
- Recursive Bayesian Updates - Implements the recursive prediction and update cycle fundamental to Bayesian state estimation.
- Kalman Filter Implementations - Provides a comprehensive library of linear and non-linear Kalman filter implementations for dynamic system state estimation.
- State Estimation Libraries - Computes probability distributions of system states using discrete Bayes and generalized filters.
- Probabilistic Estimation Toolkits - Provides a toolkit for calculating probability distributions and hidden states using discrete Bayes and generalized filters.
- Dynamic System Modeling - Merges noisy measurements with mathematical evolution models to track the state of dynamic systems.
- Hidden State Estimation - Implements Bayesian methods to estimate the internal states of dynamic systems using behavioral models and noisy data.
- Non-Gaussian State Estimation - Implements routines using Markov Chain Monte Carlo and resampling to estimate states in non-Gaussian environments.
- Non-Linear State Estimation - Tracks systems with non-linear dynamics using particle filters and Markov Chain Monte Carlo methods.
- Particle Filter Sampling - Implements particle filtering using discrete samples and resampling to handle non-Gaussian state distributions.
- State Covariance Modeling - Uses linear algebra and covariance matrices to track the mean and uncertainty of a system state over time.
- Particle Filtering - Estimates states in non-Gaussian environments using Markov Chain Monte Carlo routines and resampling.
- Discrete Bayes Filtering - Calculates probability distributions over a discrete set of states to identify the most likely current state.
- Taylor Series Approximations - Approximates non-linear system dynamics using Taylor series expansions to maintain Gaussian state distributions.
- Sensor Noise Filtering - Implements generalized structures for reducing noise and providing critical damping in raw sensor signals.
- Fixed-Lag Smoothing Algorithms - Implements fixed-lag smoothing to refine historical state estimates using a window of future data.
- Least Squares Estimators - The project derives the most probable state of a system using least squares and fading memory filters.
- Dynamic System Simulators - Generates noise matrices and simulates uncertainty within physical systems for algorithm validation.
- Statistical Analysis Libraries - Provides a framework for Gaussian analysis using Mahalanobis distance, log-likelihood, and covariance ellipses.
- Likelihood Evaluation - Evaluates the quality of state estimates by calculating the log-likelihood of observed measurements.
- Time Series Signal Smoothing - Reduces noise in sensor data using backward smoothing and fading memory filters to refine state estimates.
- Time Series Smoothing - Refines previous state estimates through fixed-lag and backward smoothing techniques.
- Time Series Smoothing Libraries - Ships a suite of algorithms for noise reduction via fixed-lag and backward smoothing.
- Time Series Analysis - Library for Kalman filtering and optimal state estimation.
