Material Insights · Understanding the Material World

State estimation (Kalman filter)

Combining a model with noisy measurements gives the best estimate of where a system is.

Mathematics and information · Control theory · No. 129

  • Electronic age
  • Science
Mathematics and information · The science behind technology

A 1960 method for blending a model with noisy readings was used on all six Moon landings.

Chart: Illustration: State estimation (Kalman filter)

Where it comes from

Chart: Map of where it comes from

Rudolf Kalman, a mathematician in Baltimore, published the filter in March 1960. That autumn he visited Stanley Schmidt at NASA Ames, who saw that it could solve Apollo's problem of knowing where a spacecraft was.

Three things to know
  1. Kalman's 1960 paper appeared in a mechanical engineering journal, not an electronics one.
  2. The method is best for systems that can be written as equations with bell-curve noise; harder cases need extended versions.
  3. An extended form of the filter was loaded on Apollo 11, whose guidance computer had less than one megabyte of memory.
The journey
  1. 1 ProblemSensors give noisy readings of where a craft is
  2. 2 ModelEquations predict how it should move
  3. 3 FilterKalman blends the two, 1960
  4. 4 ApolloNASA Ames turns it into a working navigation tool, 1961
  5. 5 TodayAircraft flight control, spacecraft navigation and driverless cars
Go deeper pick a tab to open it in full