Hands-On Delivery Routes Optimization (TSP) with AI, Using LKH and Python | by Piero Paialunga | January, 2025

The code for this article can be found in this GitHub folder.

Oone of my favorite professors in my studies told me this:

“Just because your algorithm isn't working well, doesn't mean there's a problem hard“

This means that if you want to solve any problem (simple or difficult), there will always be a way that is not reasonable enough to not work very well. For example, let's say you have to go to work at a new workplace. Instead of using Google Maps, you start on the road to your house and try all possible road combinations (north, south, west, and east). By the time you get to work, your company may have filed for bankruptcy or fired you.

Let's try to be a little more organized. Let's say in any business or engineering field, you have to find out minimum or the highest of work. For example, your company should increase revenue from sales of a certain department. We call this work f. The “strings” you pull, that is decisions which you can take to increase your income is a vector x. You obviously can't…