Planning system for deliveries in Medellín

Here we present the implementation of an application capable of planning the shortest delivery route in the city of Medellín, Colombia. We discuss the different approaches to this problem which is similar to the famous Traveling Salesman Problem (TSP), but differs in the fact that, in our problem, we can visit each place (or vertex) more than once. Solving this problem is important since it would help people, especially stores with delivering services, to save time and money spent in fuel, because they can plan any route in an efficient way. To solve this we need to construct a subgraph with the delivering points, based on the city’s map, and it will be a complete one i.e. all of its vertices are connected. Then we will give the user different options that will determine which algorithm will be used to solve the problem. Between these options there is only one that will surely give the shortest route and works only with twenty or less points. The other options are quite fast but may or may not give the shortest route. Depending on the chosen algorithm, the results in time, memory and total distance will vary. For example, we have an algorithm that does not create a subgraph to give an answer, so it takes less memory and time, but will not give the total distance. Others can give a better answer quite fast, even though they require to compute a subgraph, but still the tour produced may not be the shortest one. At last, there is an algorithm that can give the shortest route every time, but needs to look through all possible answers so it takes much more time. For the problem of planning delivery routes in Medellín our proposed solution to find the shortest route can be of huge help for small companies if their couriers do not visit more than 20 points per trip.