Independent learners in abstract traffic scenarios
Visualizar/abrir
Data
2012Tipo
Abstract
In transportation systems, drivers usually choose their routes in an uncoordinated way. In general, this leads to poor global and individual performance regarding travel times and road network load balance. This work presents a reinforcement learning based approach for route choice which relies solely on drivers’ experience to guide their decisions. There is no coordinated learning mechanism, thus driver agents are independent learners. Our approach is tested in two abstract traffic scenarios a ...
In transportation systems, drivers usually choose their routes in an uncoordinated way. In general, this leads to poor global and individual performance regarding travel times and road network load balance. This work presents a reinforcement learning based approach for route choice which relies solely on drivers’ experience to guide their decisions. There is no coordinated learning mechanism, thus driver agents are independent learners. Our approach is tested in two abstract traffic scenarios and it is compared to other route choice methods. Experimental results show that drivers learn routes in complex scenarios. Moreover, the approach outperforms the compared route choice methods regarding drivers’ travel time. Also, satisfactory performance is achieved regarding road network load balance. The simplicity, realistic assumptions and performance of the proposed approach suggest that it is a feasible candidate for implementation in navigation systems for guiding drivers decisions regarding route choice. ...
Contido em
Revista de informática teórica e aplicada. Porto Alegre. Vol. 19, n. 2 (2012), p. 13-33
Origem
Nacional
Coleções
-
Artigos de Periódicos (39166)Ciências Exatas e da Terra (5964)
Este item está licenciado na Creative Commons License