Prediction of Pedestrians' Destinations based on the Theory of Probabilistic Projection Space
Download as PDF
DOI: 10.23977/icasit.2019.001
Author(s)
Guangyao Zhou, Zhaoxin Chen, Yang Liu, Wanli Dang, Ye Pan
Corresponding Author
Guangyao Zhou
ABSTRACT
Pedestrian motion model is a research focus, involving a variety of realistic scenarios, such as transportation hub, shopping mall construction and evacuation, urban planning and large-scale event risk prevention. Its related multi-disciplines include physics, mathematics, psychology and management. The uncertainty of motion from pedestrian autonomy and complexity of real life scenes make it difficult to quantitatively calculate the movement of pedestrian, which limits our ability to establish pedestrian movement model. In current, there is not any universal model for describing pedestrian movement and for describing the impact of environment and pedestrian psychology on pedestrian movement. In paper, I use the theory of probability field and probabilistic projection space to establish a theoretical model of pedestrian movement in scenes with obstacles and destinations and to predict the destinations of pedestrians. Experimental results of a realistic scene verified that this model can accurately predict the destinations of pedestrians in general scenarios. I expect this descriptive model of pedestrian movement in complex space based on the theory of probability field and probability projection space can play a guiding role in calculation pedestrian motion and pedestrian management.
KEYWORDS
Pedestrian Dynamic, Probability field, Probabilistic Projection Space, Field Theory, Prediction, Destination