[Playerstage-users] Best way to read the real position, and Markov localization with occupation grid

[Nicolau W. <nwe...@gm...>]

Greetings, list.  I am a new user of player/stage, and I'm taking the
first steps to build my first system.

As I imagine it happens in most cases, in my simulations I would like
to have access to the real position of the robot in time, to analise
its behavior and to calibrate my localization techniques... But what
is the best way to obtain this data?

I have already modified one of the examples to print out the odometer
position, taken from  Position2dProxy::GetY etc, and also the
simulator values taken from SimulationProxy::GetPose2d. Is there
another, better way?

For example, I've read that you can configure a Position2dProxy to
deliver the true coordinates (I believe it's what is called "GPS" in
the configuration files). Can I have two Position2dProxys in my robot,
one giving me the odometer reading, and other the true position?...
And would a position proxy have any advantages or disadvantages over
using the simulation proxy to give me the real pose?...

And I also heard that you can log the activity in stage. Is there any
examples of how to do that to analyze an algorithm? (I imagine this
keeps the client programs cleaner, and helps to centralize the data
for analysis?... )

I am building a player/stage system (simulating a pioneer robot) to
implement Markov localization with an occupation grid. I imagine this
has been done many times before, does anybody has any code that could
help me?

Thanks so very much!... I'm pretty sure there are more things in
Player/Stage then are dreamt of in my philosophy!... :)

      ++nicolau




-- 
Nicolau Werneck <nwe...@gm...>         9F99 25AB E47E 8724 2F71
http://www.lti.pcs.usp.br/~nwerneck                   EA40 DC23 42CE 6B76 B07F