Efficient Gaussian Mixture Filters based on Transition Density Approximation The work was partially supported by the Ministry of Education, Youth and Sports of the Czech Republic under project ROBOPROX - Robotics and Advanced Industrial Production CZ.02.01.01/00/22_008/0004590.

Ondřej Straka, Uwe D. Hanebeck, Intelligent Sensor-Actuator-Systems Laboratory (ISAS) Karlsruhe Institute of Technology Karlsruhe, Germany E-mail: [email protected] · 2025

Read Efficient Gaussian Mixture Filters based on Transition Density Approximation The work was partially supported by the Ministry of Education, Youth and Sports of the Czech Republic under project ROBOPROX - Robotics and Advanced Industrial Production CZ.02.01.01/00/22_008/0004590. by Ondřej Straka, Uwe D. Hanebeck, Intelligent Sensor-Actuator-Systems Laboratory (ISAS) Karlsruhe Institute of Technology Karlsruhe, Germany E-mail: [email protected] (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-20002.txt.

Description

Ondřej Straka and Uwe Hanebeck keep a tracking filter's mixture from exploding in a 2025 paper. Filters that follow nonlinear systems often carry a mixture of Gaussian guesses. The standard moves multiply components at every step, forcing brutal pruning that costs accuracy and time. Pre-digesting the noise into mixtures only worsens the growth.

Their idea is to approximate the transition density itself, the law that carries the state forward, as a tidy axis-aligned Gaussian mixture. Two routes to that approximation are derived. Because the transition is approximated once, the posterior keeps a constant number of components forever, with no explicit reduction step. The design also exposes a clean dial: more components buy estimation quality, fewer buy speed. The authors position the method for tracking problems where mixture filters were previously too heavy to run in real time.

Catalog record

CREATOROndřej Straka, Uwe D. Hanebeck, Intelligent Sensor-Actuator-Systems Laboratory (ISAS) Karlsruhe Institute of Technology Karlsruhe, Germany E-mail: [email protected]
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-20002

Browse the full library · About the Tower of Babel