Now, I am trying to use Pathfinder to analyse evacuation process, expecially, I give a lot of attention to agent's complex cognitive behaviors, such as, agents will slow down speed when blocked by smoke and may replan the route to the safer area in fire emergency. But I don't know how to customize these cognitive rules within Pathfinder software. I want to know whether Pathfinder could serve to achieve this goal, if not, could you please recommend me some other tool that may meet my need? Thanks!
Dear Liang,
I have made a first attempt to link FDS data with Pathfinder trajectories as you can read in this paper:
https://www.researchgate.net/publication/303832899_A_dynamic_approach_for_the_impact_of_a_toxic_gas_dispersion_hazard_considering_human_behaviour_and_dispersion_modelling
As I know at this stage you can not customize cognitive rules within Pathfinder. However you can do it using other platforms such as Unity: https://unity3d.com/
Please contact me by email for more details: [email protected]
Regards
Ruggiero
Article A dynamic approach for the impact of a toxic gas dispersion ...
Dear Ruggiero,
thanks for your answer. I have read your paper but I am a little confused about how your integrate real-time data produced by CFD algorithm with cognitive agents. Have your considered the impact of toxic gas on human behavior,? For instance, pedestrians may slow down and replan route choice when blocked by dense toxic gas.
As you mentioned in the paper, the coupling of the impact of toxic gas on human behavior, such as, pre-evacuation behavior, way finding, is a very tough job. I agree with this very much. I am not clear of whether your have considered this tough task in your model, may be under some assumptions?
Best regards
Liang
Dear Liang,
in line with the suggestion of Peng, I would add that you need to download the entire FORTRAN code of FDS+Evac, modify it and finally compile your new version. You can find the full code here: https://github.com/firemodels/fds-smv
In my paper I could not include the impact of smoke and fire on evacuee pre-evacuation behavior and movement speed as it was not possible to include those behavioral rules in Pathfinder.
Finally you might be interested on those papers showing how fire and smoke can impact
1. pre-evacuation decision-making
https://www.researchgate.net/publication/301749700_An_Evacuation_Decision_Model_based_on_perceived_risk_social_influence_and_behavioural_uncertainty
https://www.researchgate.net/publication/282332728_A_model_of_the_decision-making_process_during_pre-evacuation
2. route choice
https://www.researchgate.net/publication/304038281_A_Mixed_Logit_Model_for_Predicting_Exit_Choice_during_Building_Evacuations
https://www.researchgate.net/publication/236593706_Movement_speed_and_exit_choice_in_smoke-filled_rail_tunnels
3. evacuee speed
https://www.researchgate.net/publication/236593706_Movement_speed_and_exit_choice_in_smoke-filled_rail_tunnels
Article An Evacuation Decision Model based on perceived risk, social...
Article A model of the decision-making process during pre-evacuation
Article A Mixed Logit Model for Predicting Exit Choice during Buildi...
Article Movement speed and exit choice in smoke-filled rail tunnels
Hi all,
In line with the suggestions of Ruggiero and Peng, I would like to add that the choice of the model depends on what are you expecting to do? What is the purpose of your research?
Pathfinder doesn't use any behavioural rules. All the behaviours should be custumised by the user.
FDS+Evac includes some behaviours (herding behaviour, following, etc.).
For pahfinding, both models suppose that the occupants know (or are familiar) at least the position of the main exit which is not realistic. If you want to simulate the cognitive rules, I think, you have to understand first all the assumptions made by the developpers and then modify the source code. Pathfinder is a blackbox, so you cannot modify it. FDS+Evac is an open source but it is in FORTRAN :S.
Another solution, is to develop your own model.
If you need more detail, feel free to contact me.
Good luck
Anass
Dear all
We surely have a good discussion here.
First, thanks for Ruggiero's warm recommendtions. I will definitely read your paper.
Before starting this discussion, I even donnot know that FDS+Evac is compiled by Fortran which is maybe a barrier when we try to costomize to our own purpose. It seems that we have drawn a conclusion that if we want to integrate complex cognitive behaviors into our agent based models, we should code from the very begining which may lack verification and validation. I find a code library PEDSIM at http://pedsim.silmaril.org/ which is a social force model framework.
I am not sure whether I get the right idea about this problem.
Looking forward to hearing from all of you.
Regards,
Liang
Dear Liang,
another option for crowd simulations is the navmesh of Unity. Have a look at those examples:
https://www.youtube.com/watch?v=C9U2XQOAUTY
https://www.youtube.com/watch?v=jtRwr3GWXYU
https://www.youtube.com/watch?v=bOKuoCLSA48
https://www.youtube.com/watch?v=9t6dDctu990
In this case the agents use the Unity navmesh to move
https://docs.unity3d.com/Manual/nav-BuildingNavMesh.html
and you can define their behaviours by using scripts in C# or Javascript. Anass can provide more info on Unity as he has been working on it here at the University of Auckland.
Regards
Ruggiero
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling