An evolutionary algorithm for the vehicle relocation problem in free floating carsharing

Abstract

In this paper, we propose a novel algorithmic solution to the vehicle relocation problem in free floating carsharing systems. In this type of systems, a set of vehicles is distributed in a city and made available for customers. After a customer rents a vehicle for a period of time, he/she returns it at any place within the operation area of the vehicle operator. Such type of carsharing systems can quickly become imbalanced in a sense that many vehicles might be left in areas with low customers demand while high-demand areas might contain very few vehicles. Therefore, a relocation process is required to retain the balance by transporting vehicles from low to high-demand areas. The relocation is done by a set of workers sharing a shuttle. We consider maximizing the number of relocated vehicles, that can be done within a given time frame, and minimizing the travel duration of the shuttle. The problem is modeled as a generalization of the pickup and delivery problem which is an NP-hard optimization problem. To solve the problem, we propose an evolutionary algorithm which has been tested on real world problem instances under different settings. Experimentation results showed that the algorithm was able to successfully solve the problem and cope with the different problems settings in reasonable time.