Refining Wii accelerometer data with Kalman filtering

Wednesday 20th June 2007, 07:09:00 PM, written by Farid

For the vector mathematics fans out there working on Wii games, PC homebrew with accelerometers or even the PS3 SIXAXIS, Gamasutra has published an article on improving the input data results from small accelerometers, such as those found in the Wii controller or the PS3 one, by using a Kalman filter model.

The Kalman filter is used, in numerous fields, to filter accurately vectors in space and thus ameliorate position and velocity estimations. And according to the author, Bertis Rasco, this method can be implemented on the Wii and ameliorate the motion detection capabilities of the machine quite visibly. The author offers a caveat in the sense that he warns that complex models will require large memory footprint and due to the instantaneous and ever changing nature of a video game, sampling will have to be done many times per second and thus it will be computational heavy, something the Wii PPC-based CPU, Broadway, should have a hard time to cope with.

The obviously math heavy article propose an overview of the filter as well as a short examples (source code included).
Discuss on the forums

Tagging

wii ± accelerometers, kalman, filter

Related wii News

E3 2008: Nintendo Press Conference Highlights
Wii get(s) 1:1 motion control
Wii reaches 5 million units threshold in Japan
Carl Bender's Nintendo's Wii: more than the sum of its parts (editorial)
NewsWeek, with a little help, takes on Wii as Gamecube 1.5
Gamebryo 2 Engine now available for the Wii
Wii outsells PS3 6-to-1 in Japan in June, 5-to-1 in May, 4-to-1 in April
Nintendo to sell new downloadable games
Nintendo's market capitalisation entered Japanese Top 10 for a brief while
Speech recognition middleware VoiceIn now available to Wii developers