A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking

By M. Sanjeev Arulampalam, Simon Maskell, Neil J. Gordon, Tim Clapp. IEEE Transactions on Signal Processing, 2002

12,450 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/7f0bbe9dd4aa3bfb8a355a2444f81848b020b7a4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

Citations in the corpus

Power System Dynamic State Estimation: Motivations, Definitions, Methodologies, and Future Work (2019, background): “general; it cannot be determined analytically for the nonlinear dynamical system model [65].”
Localization in highly dynamic environments using dual-timescale NDT-MCL (2014, methodology): “Finally, a resampling step is needed in order to avoid particle divergence [18, 19].”
Fusion of WiFi, Smartphone Sensors and Landmarks Using the Kalman Filter for Indoor Localization (2015, methodology): “The potential sensor fusion techniques for WiFi and PDR approaches include the particle filter [29], hidden Markov model [30], Kalman filter [27], etc.”
Efficient Probabilistic Inference in the Quest for Physics Beyond the Standard Model (2018, methodology): “Inference engines available in PPLs range from MCMC-based lightweight Metropolis Hastings (LMH) [90] and random-walk Metropolis Hastings (RMH) [63] to importance sampling (IS) [11] and sequential Monte Carlo [34].”
Bayesian statistical data assimilation for ecosystem models using Markov Chain Monte Carlo (2007, methodology): “Unlike SIR, the approach does not rely on resampling and consequently does not exhibit sample (ensemble) degeneracy, a problem that itself has been the subject of a great deal of research effort (Arulampalam et al., 2002; Dowd, 2006).”
Unfreezing the robot: Navigation in dense, interacting crowds (2010, methodology): “To cope with the multimodality, we use an approximate inference technique based on importance sampling, a well understood approximate inference method for Bayesian statistics (for an introduction see [1], [4], [21], [24]; for a more detailed, up to date analysis of the method see [5], [13]).”
Multi-modal estimation with kernel embeddings for learning motion models (2013, background): “However, in general these methods suffer from very poor dimensional scaling and difficulties with sample degeneracy or impoverishment [12], [4].”
Face Expression Recognition and Analysis: The State of the Art (2012, methodology): “Out of these methods, particle filters (and adaptive particle filters) are more extensively used since they can deal successfully with noise, occlusion, clutter and a certain amount of uncertainty [62].”
An Improved Exponential Model for Predicting Remaining Useful Life of Rolling Element Bearings (2015, methodology): “PF is derived from traditional filtering algorithms, such as Kalman filtering and extended Kalman filtering [23].”
Euro Area Inflation Persistence in an Estimated Nonlinear DSGE Model (2008, positive): “For concise and e¤ective introductions to these methods, see Arulampalam et al. (2002), Doucet et al. (2001).”
Structured Pruning of Deep Convolutional Neural Networks (2015, methodology): “Particle Filtering (PF) finds its applications in several fields [Arulampalam et al. 2002; Nummiaro et al. 2003; Nakamura et al. 2009; Carpenter et al. 1999].”
Transferring Rich Feature Hierarchies for Robust Visual Tracking (2015, background): “Most of the existing tracking meth-ods belong to the general framework of Bayesian tracking [1].”
An Ensemble Multiscale Filter for Large Nonlinear Data Assimilation Problems (2008, methodology): “The ensemble Kalman filter deals with nonlinearity during the forecast step by working with ensembles of randomly generated replicates represented by x jt | t 1 (for the forecast ensemble at t) and x jt | t (for the updated ensemble at t ; Arulampalam et al. 2002; Gordon et al. 1993).”
Cloud-enabled prognosis for manufacturing (2015, background): “This degeneracy implies that a large computational effort is devoted to updating particles whose contribution to the approximation of the posterior PDF is almost zero [140].”