English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Open

Presents the pdf analysis as premium feature

PDF Analysis

Insights

Presents the summarisation as premium feature

Summarisation

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

A very successful model for simulating emergency evacuation is thesocial-force model. At the heart of the model is the self-driven force that isapplied to an agent and is directed towards the exit. However, it is not clearif the application of this force results in optimal evacuation, especially incomplex environments with obstacles. Here, we develop a deep reinforcementlearning algorithm in association with the social force model to train agentsto find the fastest evacuation path. During training, we penalize every step ofan agent in the room and give zero reward at the exit. We adopt the Dyna-Qlearning approach. We first show that in the case of a room without obstaclesthe resulting self-driven force points directly towards the exit as in thesocial force model and that the median exit time intervals calculated using thetwo methods are not significantly different. Then, we investigate evacuation ofa room with one obstacle and one exit. We show that our method produces similarresults with the social force model when the obstacle is convex. However, inthe case of concave obstacles, which sometimes can act as traps for agentsgoverned purely by the social force model and prohibit complete roomevacuation, our approach is clearly advantageous since it derives a policy thatresults in object avoidance and complete room evacuation without additionalassumptions. We also study evacuation of a room with multiple exits. We showthat agents are able to evacuate efficiently from the nearest exit through ashared network trained for a single agent. Finally, we test the robustness ofthe Dyna-Q learning approach in a complex environment with multiple exits andobstacles. Overall, we show that our model can efficiently simulate emergencyevacuation in complex environments with multiple room exits and obstacles whereit is difficult to obtain an intuitive rule for fast evacuation.

Deep reinforcement learning with a particle dynamics environment applied to emergency evacuation of a room with obstacles