I'm currently working on a mobile robot's indoor localization. On the perception side, I can only rely on a 2D lidar and wheel odometry.

I have used these sources as input of different localization's packages such as slam_toolbox, an icp based package that does scan-to-map matching and a landmark detector. They all give interesting outputs but for the landmark detector and the scan-to-map matcher, the results are sporadic.

I would like to fuse all the estimated positions into a more reliable one and therefore tried to use robot_localization's ukf implementation to do so. For those familiar with ros, I am trying to find the tf between odom and map based on the results of all these packages.

The problem is, all these results are based on cross-correlated measurements and this might affect the convergence of the filter.

Is it possible to fuse results based on the same input sources ?

If yes, is ukf a right way to do so ?

Otherwise, are there other filters more adapted to this application ?

Thank you in advance for your help !


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.