I think ant colony is more robust than taboo search. but algorithms based on reward such as Q-learning are more appropriate. if you want implemet new method, i think, combine ant colony taboo search model is fast in finding best solution and more robust of single ant colony.
If your are talking about vehicle routing problem, the current state-of-the-art is memetic algorithm, you can search online for algorithms such as bone-route and Active Guided Evolution Strategy (AGES), which have been shown to excel over ACO and TS.
Hi Ali, Below is the list of books that i would recommend:
For the fundamental, you may want to start with this book on the VRP chapter: "Local Search in Combinatorial Optimization". For the classical VRP, i also recommend this book "The Vehicle Routing Problem". It is classical but a bit old
For more recent advances on VRP, you can read: "Bio-inspired Algorithms for the Vehicle Routing Problem" and "The Vehicle Routing Problem: Latest Advances and New Challenges"
One of the oldest, and perhaps also the most popular algorithm for solving this problem is the "Clarke and Wright" heuristic, also called the savings heuristic. It works extremely fast, and gives quite good solutions. Many commercial software packages for solving VRP use this algorithm, as it is very easy to understand and code.
The vehicle routing problem is very well researched. TABOO search used to be the best, meaning it achieves the best results in the shortest amount of time. Over the last 5 years a new algorithm seems to have beaten it: Adaptive Large Neighborhood Search (ALNS), invented in 2006 by Stefan Røpke for Vehicle Routing. I have attached
ALNS has also been used for Pick up and delivery problems, see attached paper
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication