Here is a great article about why computerized routing programs frequently generate "optimal" results that folks don't understand, and thus "aren't good": http://www.optaplanner.org/blog/2013/08/06/FalseAssumptionsForVRP.html
Shortest routes: Not always best. Fewest trucks: Not always best. Focus on time slots before capacity: not always best. No crossing routes: not always best.
Good route results usually look something like this:
This is why NP-hard problems (like routing) are **hard**. :)
Shortest routes: Not always best. Fewest trucks: Not always best. Focus on time slots before capacity: not always best. No crossing routes: not always best.
Good route results usually look something like this:
This is why NP-hard problems (like routing) are **hard**. :)
No comments:
Post a Comment