[m-dev.] ICFP Contest: new planning idea

[Peter Moulder]

Incidentally, the first approximation I was thinking of was to manually
trace a polyline in xfig over the xpm, and parse the .fig file.

I think in some tracks this gives better paths than manhattan
distance ("path" in the sense "which side to go around an obstacle",
i.e. the choice aspect of the continuous approximation of the problem).
For the not-quite-manhattan metric I used for checking correctness of my
C++ program, the car track was such an example; I don't know whether or
not this is also true for Manhattan distance, but I'd guess it to be
true of some of the later tracks (lots & lots of obstacles).

I've done such an xfig drawing for one of the tracks, but haven't
written code to parse the resulting .fig file.  .fig files are very
parsable, btw.

I mention this alternative mainly for the case that it's considered easier
to implement than the pure-manhattan idea.  As I say, at the moment I'm
more interested in just getting solutions than getting good solutions.

"does better at the choice part of the problem" is irrelevant unless we
had time to implement refinement algorithms, and at the moment I don't
think we will do so.

pjm.

--------------------------------------------------------------------------
mercury-developers mailing list
Post messages to:       mercury-developers at cs.mu.oz.au
Administrative Queries: owner-mercury-developers at cs.mu.oz.au
Subscriptions:          mercury-developers-request at cs.mu.oz.au
--------------------------------------------------------------------------