we've got a two wheeled differential robot that shall be able to navigate in a static 2D environment. I've set up /tf to work with the location data that our sensors provide. The motors that drive the robot are steppers that don't get just the rounds per minute but additionally a step count.
Location is determined with an external triangulation system. Location and shape of all static obstacles are pre-defined, hence there's no need to generate a map at runtime. However there are dynamic obstacles in the area of operation which might be other robots or items of given shapes. Other (moving) robots shall be avoided, other non-robotic dynamic obstacles might be of interest for our robot, so collision with them isn't necessarily bad, has to be decided individually in any particular case.
We don't have a laser scanner. In my opinion we won't need one since the environment and locations and shapes of the important obstacles are known.
Is there a way to use the navigation stack anyway? Is it even necessary or are there other (easier) approaches for navigation that work in an environment such as ours?
Post score: 1
Comment by Hendrik Wiese on 2013-07-11:
I've got to add: the location of other robots is determined the same way as ours. They've got the same system mounted that does triangulation and sends its information to our robot over a ZigBee network.