Learning Automata-Based Solutions to the Multi-Elevator Problem

Abstract

In the last century, AI has been the topic of interest in many areas, where the focus was on mimicking human behaviour. It has been researched to be incorporated into different domains, such as security, diagnosis, autonomous driving, financial prediction analysis and playing games such as chess and Go. They also worked on different subfields of AI such as machine learning, deep learning, pattern recognition and other relevant subfields. Our work in a previous paper [1] focused on a problem that has not been tackled using AI before, which is the elevator-problem. In which we try to find the optimal parking floor for the elevator for the single elevator problem. In this paper, our work extends the model by solving the more complicated scenario, which is the multi elevator problem (MEP) using Learning Automata (LA). This problem can then be generalized to be applied to a variety of problems that share the same characteristics with the elevator problem. We refer to these problems as Elevator-Like Problems (ELPs). For the extended version (MEP) we try to find the optimal parking floors for the set of elevators so as to minimize the passengers’ Average Waiting Time (AWT). Apart from proposing benchmark solutions, we have provided two different novel LA-based solutions for the multi-elevator scenario. The first solution is based on the well-known LRI scheme, and the second solution incorporates the Pursuit concept to improve the performance and the convergence speed of the first solution, leading to the P $$ L_{RI} $$ scheme. The simulation results presented demonstrate that our solutions performed better than those used in modern-day elevators, and provided results that are near-optimal, yielding a performance increase of up to 91%.