Quote:
Originally Posted by JamesBrown
It is pretty complicated, it may seem fairly simple but in practice it isn't. It is even more difficult since in FIRST we can't use laser range finders. IR, and Sonar range finders introduce a lot of noise.
If you are serious about looking into this in more depth research SLAM ( Simultaneous Localization and Mapping). Nearly every University with a robotics lab is doing some level of research on SLAM algorithms, this alone should be enough evidence to show you that this is not a trivial problem.
|
For noise --- running a rolling average (16x or higher oversample) tends to level out most noise issues, although it does slow down response.
SLAM is much easier in a FIRST enviroment where the initial map is known, the objects sizes (other than other robots sizes) are known (and possibly fixed in position), the initial position and orientation (as well as size) of the robot is known, and the driving characteristics (drive train characteristics, turning radius, etc) of the robot are known. While it is not trivial ... it is doable in a FIRST enviroment.