Thanks for the help Rich. I shall get back on the steps i followed... soon.


On Mon, Nov 22, 2010 at 2:52 PM, Rich Mattes <jpgr87@gmail.com> wrote:
On 11/19/2010 08:16 PM, Paddy wrote:
Hello All,

           I am having a segmented  data (x,y,z) corresponding to the floor (indoor environment) in front of the robot from Swissranger SR-3000 sensor. I would like to know how can i use this data to plan a path from point A to B (Assuming both are on the same segment ) using Player. 

I am not sure in what way i should provide the data such that the player can plan a path accordingly. 

I am familiar with vfh, amcl, wavefront drivers and have written some basic codes in simulation (player-stage), but the simulations uses Sick-LMS laser scanner as the input sensor.
I would like to achieve similar results using the real data from SR-3000.

Your suggestions regarding the implementation steps would be highly regarded,

Thanks,
Paddy 
--
Hi Paddy,

The first way I can think of to achieve what you want to do is to somehow create a gridmap from your swissranger measurements.  You'll have to be able to look at your measurements, find where the obstacles are on the floor, and translate the obstacle positions into real (x,y) positions from your robot.

If you can reliably convert swissranger scans into (x,y) coordinates, you might be able to create a driver that generates a fake laser scan, which looks at the swissranger data and creates a laser "hit" where a laser would hit an obstacle (somehting on the floor, wall, etc).  With this synthesized laser scan, you could generate an occupancy grid using something like mricp, and then use the amcl/wavefront/vfh stack with almost no modification.

Of course, this isn't the only way to do things: you could roll your own map generator, eschew global planning and use vfh directly, etc.  Hopefully it gives you some ideas on how to move forward.

Rich



--
-Prasad