Saturation likelihood field for AMCL

asked 2019-07-02 03:43:23 -0600

graziegrazie gravatar image

updated 2019-07-03 15:32:13 -0600

I am developing localization algorithm. I'd like to compare its performance with AMCL, but I'm not sure how the parameter should be defined. Then I checked some localization related papers which use MCL for performance comparison.

[1] says "we used 5000 particles, as measurement model we used likelihood fields saturated to 2m of maximum distance, and rendered the map at 1cm of resolution for maxi-mum accuracy."

[2] says "We fine-tuned the parameters of our implementation of MCL (250mm saturation and 5000 particles) to be able to track the robot even in a highly occluded scenario."

I know MCL is different to AMCL, but quite similar. About the number of particles, I should set min_particles and max_particles as 5000. On the other hand, I'm not sure how to realize saturation likelihood about distance. Should I set laser_likelihood_max_dist as proper value? Or other parameter?

edit retag flag offensive close merge delete