In-building evacuation simulation method and system based on shared deep reinforcement learning

Abstract

The invention discloses an in-building evacuation simulation method and system based on shared deep reinforcement learning, and the method comprises: enabling a leader of each group of to-be-evacuatedcrowd to be connected with a corresponding navigation Agent, and enabling each navigation Agent to be connected with a management Agent; guiding each group of evacuation by each navigation Agent, andstoring information acquired in real time and a self experience pool into a knowledge base managed by the management Agent by each navigation Agent; learning, by the management Agent, evacuation information of all the navigation Agents in the knowledge base on the basis of a shared deep reinforcement learning algorithm, and sending a strategy generated by learning and used for guiding the navigation Agents to perform path selection to the navigation Agents in real time; carrying out path planning by each navigation Agent according to the received strategy used for guiding each navigation Agent to carry out path selection; and guiding, by the leader, the corresponding group to evacuate to the evacuation exit according to the path planning of the corresponding navigation Agent.

Description

technical field [0001] The present disclosure relates to the technical field of crowd evacuation simulation, in particular to a simulation method and system for building evacuation based on shared deep reinforcement learning. Background technique [0002] The statements in this section merely mention background art related to the present disclosure and do not necessarily constitute prior art. [0003] In recent years, with the rapid development of social economy, various large-scale public activities that gather a large number of people have been increasing, and the safety issues have attracted more and more attention. In some densely populated public buildings, due to unreasonable architectural design and inadequate emergency management, when various emergencies occur and people need to evacuate urgently, it is easy to cause crowding and obstruction of evacuated people , and even serious stampede accidents. [0004] Crowd stampede accidents are regular. Different from na...

AI Technical Summary

Problems solved by technology

[0008] Traditional reinforcement learning methods cannot solve the problem of dimensionality disaster caused by too large state space;

[0009] The traditional simulation method of crowd evacuation in buildings is slow and poor in stability, which cannot meet the needs of fast and accurate crowd evacuation simulation.

Images