Particle Filters

Ravi, I've done extensive research with particle filters. I did my Master's Thesis on image processing and video target tracking with them. They are great with nonlinear non Gaussian systems and state models..Very Very computationally intensive. Most embedded systems dont have enough power to run a good PF, or AuxPF. Even host PF applications can run slow..I guess it depends on your objectives, like what are you estimating or tracking and how fast do you need it to execute..Most of the time you can get by with the plain vanilla KF or EKF with embedded systems. EKF can use nonlinear state model with Gaussian noise assumptions where the nonlinearity can be linearized with first order Taylor series.

Replies

Colin Johnston February 18, 2012 at 10:32am

Ravi, I've done extensive research with particle filters. I did my Master's Thesis on image processing and video target tracking with them. They are great with nonlinear non Gaussian systems and state models..Very Very computationally intensive. Most embedded systems dont have enough power to run a good PF, or AuxPF. Even host PF applications can run slow..I guess it depends on your objectives, like what are you estimating or tracking and how fast do you need it to execute..Most of the time you can get by with the plain vanilla KF or EKF with embedded systems. EKF can use nonlinear state model with Gaussian noise assumptions where the nonlinearity can be linearized with first order Taylor series.

This reply was deleted.

Discussions

Particle Filters

Replies

Activity