A solution for the Elevators Group Dispatch by Multiagent Reinforcement Learning

• Jordão Memória Universidade Estadual do Ceará
• José Maia Universidade Estadual do Ceará

DOI: https://doi.org/10.5753/eniac.2019.9322

Resumo

In this work, a modeling and algorithm based on multiagent reinforcement learning is developed for the problem of elevator group dispatch. The main advantage is that, along with the function approximation, this multi-agent solution leads to reduction of the state space, allowing complex states to be addressed with a synthesizing evaluation function. Each elevator is considered an agent that have to decide about two actions: answer or ignore the new call. With some iterations, the agents learn the weights of an evaluation function which approximate the state-action value function. The performance of solution (average waiting time - AWT), shown varying the traffic pattern, flow of people, number of elevators and number of floors, is comparable to other current proposals reported in the literature.

Referências

Aranibar, D. B. (2009). Aprendizado por Reforço com Valores de Influência em Sistemas Multi-Agente. PhD thesis, Universidade Federal do Rio Grande do Norte.

Barrios-Aranibar, D. and Gonçalves, L. M. G. (2007). Learning coordination in multagent systems using influence value reinforcement learning. In Seventh International Conference on Intelligent Systems Design and Applications (ISDA 2007), pages 471– 478. IEEE.

Cao, L., Tian, J., and Zhang, Z. (2008a). Elevator group control system based on information fusion technology. In 2008 3rd International Conference on Innovative Computing Information and Control, pages 473–473.

Cao, L., Zhou, S., and Yang, S. (2008b). Elevator group dynamic dispatching system based on artificial intelligent theory. In 2008 International Conference on Intelligent Computation Technology and Automation (ICICTA), volume 1, pages 183–186.

Ikeda, K., Suzuki, H., Kita, H., and Markon, S. (2008). Exemplar-based control of multicar elevators and its multi-objective optimization using genetic algorithm. In The 23rd International Technical Conference on Circuits/Systems, Computers and Communications, page 701–704.

Ikuta, M., Takahashi, K., and Inaba, M. (2013). Strategy selection by reinforcement learning for multi-car elevator systems. In 2013 IEEE International Conference on Systems, Man, and Cybernetics, pages 2479–2484.

Liu, W., Liu, N., Sun, H., Xing, G., Dong, Y., and Chen, H. (2013). Dispatching algorithm design for elevator group control system with q-learning based on a recurrent neural network. In 2013 25th Chinese Control and Decision Conference (CCDC), pages 3397–3402.

R. S. Sutton and A. G. Barto (2000). reinforcement learning. In Tokyo: Morikita Publiching Co.

Russell, S. and Norvig, P. (2009). Artificial Intelligence: A Modern Approach. Prentice Hall Press, Upper Saddle River, NJ, USA, 3rd edition.

Takata, Y., Mikura, Y., Ueda, H., and Takahashi, K. (2010). Cooperative learning of bdi elevator agents. In ICAART 2010 - 2nd International Conference on Agents and Artificial Intelligence, Proceedings, volume 2, pages 172–177.

Valdivielso, A. and Miyamoto, T. (2011). Multicar-elevator group control algorithm for interference prevention and optimal call allocation. IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans, 41(2):311–322.

Valdivielso, A., Miyamoto, T., and Kumagai, S. (2008). Multi-car elevator group control: Schedule completion time optimization algorithm with synchronized schedule direction and service zone coverage oriented parking strategies. In The 23rd International Technnical Conference on Circuits/Systems, Computers and Communications, Session H5 Elevator Control Systems, pages 689–692.

Watkins, C. J. C. H. and Dayan, P. (1992). Q-learning. Machine Learning, 8(3):279–292.

Xue, L. H. (2002). Fuzzy neural network based elevator group control method with genetic algorithm. Master’s thesis, Tianjin University.