Revision history [back]

There is zero expectation with SLAM Toolbox that when using the localization mode you don't have a pretty good estimate of your initial pose in the first place. It is not a filter approach and was not designed for that (e.g. it was meant for professional applications with tuned platforms in deployed settings where you can persist information like last known pose when at a dock or something).

I don't know if you meant that package specifically, but you tagged that and didn't mention a particular approach in your question.

A particle filter approach may or may not do that, depending on the specific package in question -- but most will deal with the global localization problem.

W.r.t. work on matching current data to a map: check literature. There must be thousands of papers on that subject.