|
Title:
|
MULTI AGENT PREDICTIVE PEDESTRIAN MODEL BASED ON LOCAL AREA INFORMATION |
|
Author(s):
|
Manabu Gouko, Kazumichi Ohtsuka |
|
ISBN:
|
978-989-8704-04-7 |
|
Editors:
|
Philip Powell, Miguel Baptista Nunes and Pedro Isaías |
|
Year:
|
2014 |
|
Edition:
|
Single |
|
Keywords:
|
Pedestrian, multi agent model, prediction |
|
Type:
|
Poster/Demonstration |
|
First Page:
|
381 |
|
Last Page:
|
382 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
Human beings are capable of adapting their behavior according to changes in their environment. This ability of adaptation is supported by prediction of environmental dynamics. In this paper, we focused on the effectiveness of prediction to generate actions, and then developed a multi agent model in which agents decide their own behavior based on prediction of their surroundings. In the proposed model, the agents predict crowdedness of their surroundings, and they decide their own direction of movement based on this prediction. We also considered the limited cognitive load of real human beings; thus, the agents observe only a limited area in front of them, and they only observe whether other agents exist in the limited domain. We carried out simulations using our model to confirm its effectiveness. The results show that the model is able to reproduce lane formation, one of the fundamental characteristics of real world pedestrian dynamics. |
|
|
|
|
|
|
