RouteArt: A New Framework for Vehicle Routing Problem with Pickup and Delivery using Heuristic Bubble Algorithm Akin Ilker Savran 1,2 , Erhan Musaoglu 2 , Mehmet Fatih Yuce 2 , Engin Yesil 1 1 Istanbul Technical University, Control and Automation Engineering Department, Maslak, TR-34469, Istanbul, Turkey 2 LA Software Group, Cumhuriyet Caddesi Ozkan Sokak No: 2/7, Kavacik, Istanbul, Turkey savran@itu.edu.tr, erhanm@la.com.tr, mehmety@la.com.tr, yesileng@itu.edu.tr Abstract—In this study, a new framework for the vehicle routing problem with pickup and delivery (VRPPD) called RouteArt is presented. Furthermore, the enhanced Heuristic Bubble Algorithm (HBA), which is a nature-inspired algorithm, is presented as a new solution approach to the VRPPD problem. This flexible web-based framework gives the user opportunity to solve the routing problems using HBA and report the solution. In order to show the benefit of using RouteArt with HBA, five case studies are discussed, and the profit of using the proposed enhanced HBA is shown with comparisons. I. INTRODUCTION Transportation is a very significant issue in logistic distribution and supply chain management based on road system delivery. Thus, the Vehicle Routing Problem (VRP) has been mainly investigated for many years. Several proper implementations of VRP are supplied using the expansion of optimization and progressive logistic systems in order to give solutions to the real world situations [1]. The VRP was first presented by Dantzig and Ramser in 1959 [2]. Afterward, the Capacitated Vehicle Routing Problem (CVRP), the Distance Constraint Vehicle Routing Problems (DCVRP), the Vehicle Routing Problem with Backhauls (VRPB), the Vehicle Routing Problem with Time Windows (VRPTW), the Vehicle Routing Problem with Pickup and Delivery (VRPPD) are presented as an expansion of the VRP [3]. In VRPPD, the vehicles are needed to deliver and pick orders at customer positions [4]. Initially all delivers are completed, then pickups are accomplished in the customer positions (addressed) [5]. The major point needed to be set in VRPPD is the preservation of enough empty space in the vehicle to place the given back orders. This situation makes the problem more difficult since the use of the vehicle volume is incompetent, the travel distances rise and the supplementary vehicles are required. Since the VRPPD generalizes the capacitated vehicle routing problem (CVRP), the VRPPD is NP-hard in the strong sense [6]. The VRPPD has been used specially by the firms in the reverse logistics context which is the assignment of handing the reverse flow of final goods or raw materials [7]. The essential implementations of VRPPD is dial-a-ride problem (DARP), which is the transportation of handicapped and elderly people in the urban areas [8]. The aim of VRPPD is finding the optimal routes for fleet of vehicles. The optimal routes are commonly planned to mass transportation services, airlines, school bus services, library fitting and supply distributors, and grocery produce distributors [9]. The VRPPD has different variations [10] such as VRPPD with time windows [11], VRPPD with coordination of transportable resources [12], and VRPPD with finite and infinite horizon [13]. Another approach in the VRPPD is to divide the vehicles to two fleets as the pickup vehicle fleets and the delivery vehicle fleets. The transfers are introduced during the delivery phase to maintain an interaction between these two fleets [14]. A comparison of a classical VRPPD and a different model of VRPPPD where transfers are occurred among the vehicles at the depot has been studied in [15]. Additionally, another approach has been proposed in [16] that transfer places in the each positions are different from the depot. Therefore, the transfers could be occurred among the vehicles, thus, the orders are delivered to the clients by using more than one vehicle. In [17], the combined delivery and pickup problem is divided into three variations: The first variation is vehicle routing problem with backhauls (VRPB). In VRPB, all deliveries must be finished before beginning of pickup. The second variation is vehicle routing problem with mixed delivery and pickup (VRPMD), where the delivery and pickup are made with different nodes without any constraints on arrangement or many visits by a vehicle to a node. The third variations is vehicle routing with simultaneous delivery and pickup (VRPSDP). In VRPSDP, every node of visit are made delivery and pickup. Briefly, the VRPPD is to determine optimal route for vehicle fleet that deliver the consignments to delivery addresses and take away the consignments to pick up addresses [18]. The VRRPD has been studied since 1970s and many solution methods have been found as other optimization problem. One of the solution methods is exact methods are heuristic approaches and meta-heuristic approaches. The heuristic approaches are tabu search algorithm, ant colony optimization, genetic algorithm (GA), and particle swarm optimization algorithm (PSO) are usually approved to solve VRPPD [20]. Lately, a new heuristic approach called Heuristic Bubble Algorithm (HBA) is proposed as a solution to VRPPD [3]. The results obtained using Matlab simulations shows the benefit of HBA. In this study, a new VRPPD framework, which is called RouteArt, is presented. The main advantage of this web- based framework is HBA algorithm. For RouteArt, the HBA presented in [3], is extended to give better solutions. In order to show the benefit of the proposed enhanced HBA