Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Modify pedestrian dynamic model

For (ii) you might be able to hack something together with TraCI (person.setSpeed, https://sumo.dlr.de/docs/TraCI/Change_Person_State.html): whenever an S encounters a D ahead of an intersection you could slow down it's maximum speed to that of D so they will only cross together. After the the intersection you could reset the speed for S.

Am Di., 22. Nov. 2022 um 01:13 Uhr schrieb Thanhnt Nguyen <thanhnt.fit@xxxxxxxxx>:
Thank you, Mr.Jakob
About (ii), yes, my main goal is to get temporary formation of pedestrian groups. Realistic looking visualization is also important but not the main objective.
 
(iii) if it's impossible to take the length of a platoon into account, I would change a bit in my environment, for example, the platoon has a very small distance between each AGV, then no one could come between two AGVs. Therefore the platoon of several AGVs would become a unique AGV which is long, containing several carriages (each carriage is a single AGV). This long AGV lets pedestrians notice it and the length prolongs waiting time for them. Therefore, due to waiting so long, the gap acceptance of pedestrians will reduce in the near future.


Vào 22:47, Th 2, 21 thg 11, 2022 Jakob Erdmann <namdre.sumo@xxxxxxxxx> đã viết:
(i) you can define different types of pedestrians (i.e. name S,N,D, etc) and assign different maximum speeds
(ii) complex group behavior as described is quite far from the current models. What is the main goal here?
    - accurate travel times for persons (might work by making some persons slower)
    - accurate travel times for AGV (might work by making some persons "bigger")
    - temporary formation of pedestrian groups that block a bigger part of the floor space (assemble, cross, disperse)?
    - realistic looking visualization?
My guess is that the latter two points would be quite involved to model and implement.

(iii) gap acceptance in sumo currently has no way to take platoons length into account. The pedestrian behavior is using similar methods as that used by cars coming from a minor road and this only looks at the closest oncoming car. SUMO does allow some hooks to purposefully ignore conflicts and you might use this to tune the gap acceptance (https://sumo.dlr.de/docs/Definition_of_Vehicles%2C_Vehicle_Types%2C_and_Routes.html#transient_parameters)

regards,
Jakob

Am Mo., 21. Nov. 2022 um 06:03 Uhr schrieb Thanhnt Nguyen <thanhnt.fit@xxxxxxxxx>:
Thank you Mr.Jakob
Let me briefly introduce my healthcare environment: individuals move with AGV. The AGVs are only to transfer materials in the environment (a flat floor) like the following picture. AGV might meet pedestrians who include: staff (S for short), visitor (V for short), patients - move like normal (N for short) and individuals with disabilities (D for short)


Therefore I need some improvement:
(i) in XML, the personFlow needs new attributes, for example: disable (a probability indicates the person is D or not). If a pedestrian is D type, he might have a stick or crutch or wheeled-chair or even he is bind.
(ii) In the intersection, D person (who has a slower velocity) might be carried by other S, V or even N people for crossing, especially as the D is moving in a crossway. After leaving the crossway, D person could be left alone.
(iii) A pedestrian will have different gap acceptance (a distance between person to the AGV). If the distance is too short, he/she would decide to stay at the waiting area.
image.png

The gap acceptance of a pedestrian depends on yielding of vehicle (too loud or not), healthy of the person as well as the number of AGV's platoon
Looking at the code of Omnet++ and SUMO, I still don't know which files to modify. If it is too difficult to change the code, perhaps I have no choice but to use a quick and dirty hack to get my purpose.
Thank you in advance

On Sun, Nov 20, 2022 at 9:51 PM Jakob Erdmann <namdre.sumo@xxxxxxxxx> wrote:
Some comments:
- the paper ""Modelling pedestrian dynamics in SUMO" doesn't describe the whole set of vehicle/pedestrian interactions that are currently implemented
- the SUMO team is currently engaged with the integration of "social force" style pedestrian models by way of JuPedSim (jupedsim.org), https://github.com/eclipse/sumo/tree/jupedsim

Can you describe the interactions that you need to simulation in more detail?



Am So., 20. Nov. 2022 um 14:41 Uhr schrieb Thanhnt Nguyen <thanhnt.fit@xxxxxxxxx>:
Dear all
As reading the papers: "Modelling pedestrian dynamics in SUMO" and "Pedestrian Simulation In SUMO Through Externally Modelled Agents" I notice that the current pedestrian model in SUMO is very simple.
In my work, I need a pedestrian model which reflects a healthcare environment. In this model, I intend to implement interaction between vehicles (Automated Guided Vehicle) as well as individuals with disabilities
I still don't know how to implement a new pedestrian model in SUMO, could I need to modify the code of Viens?
Does anyone have a hint for me by chance?
Thank you in advance

_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user


--
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

Back to the top