I made a SLAM (Simultaneous Localization And Mapping) using Extended Kalman Filter (EKF) and it works really good, but I want to see if it works better using Neural Networks.

For the EKF I used an initialization step, prediction step and update step,how can I change that to Neural Networks?


  • $\begingroup$ I am not sure this can work in the way your question implies. You could certainly use a Neural Network as a parametric model for estimating state transitions and whatnot within an EKF framework, but a Neural Network on its own wouldn't act as a Bayesian estimator like a Kalman Filter variant does. $\endgroup$ – spektr Feb 14 '17 at 18:40
  • $\begingroup$ And how can I do that? I have no idea and I would like to know $\endgroup$ – Unnamed Feb 19 '17 at 16:43
  • $\begingroup$ @choward can you please help me? $\endgroup$ – Unnamed Feb 21 '17 at 13:51
  • $\begingroup$ you can use the EKF/UKF as a nonlinear parameter estimation tool where the weights of the Neural Network are what you are estimating. If you look into nonlinear parameter estimation using EKF/UKF, you should be able to find some references. Here's one reference I have used in the past: ieeexplore.ieee.org/document/940586 $\endgroup$ – spektr Feb 21 '17 at 14:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.