On Mon, Aug 9, 2010 at 7:58 AM, rfavier <raphael.favier@gmail.com> wrote:
> Ho! and is there a solution to solve the kidnapped robot problem?

hi Raphael,

In the literature, yes, there are many solutions to this problem.
Neither gmapping nor amcl offer a solution, however (amcl allows for
global localization, in which particles are distributed evenly
throughout the environment, but the resulting localization estimate
will depend strongly on the structure of the environment, and how long
you're willing to wait for the filter to converge).

We (at Willow) have done some work on using visual features to
recognize places (visual features are usually more distinguishing than
laser features).  But we haven't yet released a stable place
recongition system.

	brian.