A simulation method and device for crowd evacuation based on reinforcement learning

Abstract

The invention relates to a reinforcement learning-based crowd evacuation simulation method and apparatus. The method comprises the steps of obtaining a video of a specified region, and extracting crowd evacuation paths, initial coordinates, a motion speed and an exit position from the video by utilizing a KLT tracking algorithm; storing the crowd evacuation paths in a point-to-line mode; according to the initial coordinates, grouping crowds, and selecting pedestrians closest to the exit position in groups as leaders in the groups; selecting optimal exits as evacuation targets by the leaders, selecting optimal evacuation paths in the crowd evacuation paths by utilizing a reinforcement learning algorithm, and leading general pedestrians in the groups to move according to the optimal evacuation paths by the leaders; and storing the optimal evacuation paths of the groups to serve as recommended paths of an evacuation drill.

Description

technical field [0001] The invention relates to the field of crowd evacuation computer simulation, in particular to a crowd evacuation simulation method and device based on reinforcement learning. Background technique [0002] With the rapid development of my country's economy, people's living standards have been improved rapidly, and people travel more and more frequently, especially in public places with dense crowds, such as train stations, squares, shopping centers, etc. Small disturbances will have a great impact on the rapid evacuation of the crowd, and there are great potential safety hazards. If the crowd cannot be effectively controlled, it will easily lead to crowded and stampede incidents. Moreover, research also shows that in the event of an emergency such as a fire or an earthquake, if there is no reasonable guidance, the crowd will fall into extreme panic, and due to the influence of the nearby psychology, the crowd often cannot effectively use the multiple exit...

AI Technical Summary

Problems solved by technology

In the more common computer simulation models at present, the macroscopic model regards the crowd as a whole. Although the simulated evacuation speed is fast, it ignores the differences among pedestrians, so the evacuation effect is not ideal. Although the microscopic model is independent, the individual The movement only follows the rules defined by the model, lacks macroscopic target selection and path navigation, and requires global movement planning for each evacuated individual. The computational overhead is too large, and it can only be used to deal with small-scale groups. In real-time Difficult to apply in computing environment

Therefore, the existing evacuation models do not make good use of real evacuation to guide the simulation

Images