Itinerary Generation Case Study
4. RELATED PROBLEMS
The itinerary generation problem can be modelled based on any of these three problems.
• Travelling Salesman problem
• Team-orienteering problem
• Vehicle Routing problem
Solving these problems would mean solving the itinerary problem itself. Each of these problems can have additional constraints like timing window etc. which drastically change the solution space of the problem. Choosing the right problem to model upon and then choosing the right method to solve that problem is crucial to the quality of the solutions generated. We discuss each of the problems, prominent solutions to those problems and what kind of constraints they can handle.
4.1 Traveling Salesman problem
Problem statement: Given a list of cities and the distances between each pair of cities, what is the shortest possible route that visits each city exactly once and returns to the origin city? It is similar to Itinerary generation problem where-in we want the shortest route through all the POIs. Since we get only a single route on solving the TSP, it can provide only single day itineraries. TSP-TW is a variation of TSP which addresses timing window constraints.
The general class of local-search algorithms …show more content…
(2010) [17]. The MuPOPTW allows multiple time windows. Each customer may be visited on different days, on different time slots and each customer may have multiple time windows for a given day. A Variable Neighborhood Search (VNS) algorithm was introduced in order to provide good solutions in reasonable computation time. The VNS was adapted to the OPTW and TOPTW benchmark instances. VNS is able to find the optimal solution for each instance at least once over 10 runs for the OPTW instances. VNS also provides very good solutions but it requires more computation time for solving the TOPTW
The itinerary generation problem can be modelled based on any of these three problems.
• Travelling Salesman problem
• Team-orienteering problem
• Vehicle Routing problem
Solving these problems would mean solving the itinerary problem itself. Each of these problems can have additional constraints like timing window etc. which drastically change the solution space of the problem. Choosing the right problem to model upon and then choosing the right method to solve that problem is crucial to the quality of the solutions generated. We discuss each of the problems, prominent solutions to those problems and what kind of constraints they can handle.
4.1 Traveling Salesman problem
Problem statement: Given a list of cities and the distances between each pair of cities, what is the shortest possible route that visits each city exactly once and returns to the origin city? It is similar to Itinerary generation problem where-in we want the shortest route through all the POIs. Since we get only a single route on solving the TSP, it can provide only single day itineraries. TSP-TW is a variation of TSP which addresses timing window constraints.
The general class of local-search algorithms …show more content…
(2010) [17]. The MuPOPTW allows multiple time windows. Each customer may be visited on different days, on different time slots and each customer may have multiple time windows for a given day. A Variable Neighborhood Search (VNS) algorithm was introduced in order to provide good solutions in reasonable computation time. The VNS was adapted to the OPTW and TOPTW benchmark instances. VNS is able to find the optimal solution for each instance at least once over 10 runs for the OPTW instances. VNS also provides very good solutions but it requires more computation time for solving the TOPTW