Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-user] Modify position, vel and acceleration of each actor

I think this topic is quite similar to Github issue #13647 in expecting specific control over internal computation of dynamics. The main question is: Should a use of moveToXY() invalidate computation of speed and acceleration and which values sumo should return when asked via TraCI?

 

At the moment, speed and acceleration are updated on moveToXY() and undergo a series of checks to avoid extreme values. The setSpeed command sets a new speed target for the vehicle in the following time steps (until disabling it with a negative value).

 

Best regards

Mirko

 

 

 

 

-----Original-Nachricht-----

Betreff: Re: [sumo-user] Modify position, vel and acceleration of each actor

Datum: 2023-09-14T10:59:51+0200

Von: "Ruud van Gaal via sumo-user" <sumo-user@xxxxxxxxxxx>

An: "Stefano Bonasera" <stefano.bonasera@xxxxxxxxx>

 

 

 

Hi Stefano,
 
Indeed, no changing in vehicle state takes place until you simulate a step. You would say it's a linear system but I can imagine two methods:
 
1: v=v+a*dt, p=p+v*dt
2: p=v+v*dt, v=v+a*dt
 
This may account for a change in your expectations regarding position after a simulation step. What do you get for position when setting position to (0,0,0), adding velocity of (1,0,0) and an acceleration of (10,0,0) and taking a small step?
Note also that I also call Simulation::step(0) to step one-by-one. This makes it possible to set my own position/velocity (etc) at every step. If I don't and use Simulation::step(tNow), internally after 1 timestep, SUMO takes over and starts calculating its own speed etc again. So make sure you overrule your inputs at every step.
 
Cheers,
Ruud
 
 

On Wed, Sep 13, 2023 at 5:43 PM Stefano Bonasera <stefano.bonasera@xxxxxxxxx> wrote:
Hi Ruud,

Thanks for the answer.
I noticed that with your set of commands I cannot see an actual modification of actors position and velocity: this means that, if after your commands I getPosition() or getSpeed() for the "modified" actors, I will then get the same values that I had prior to your list of commands - this is why I need to simulate the step.
Sorry, the 'get the final wanted result' in my case means that I can: 
1) modify position velocity and acc of each actor that I want (say, actor 1 was at (x, y, v, a) = (1, 1, 1, 1) but I want it at (x, y, v, a) = (2, 3, 4, 5))
2) when I try to getXXX(), I would then get the values that I wanted for each specific actor (following above, I get (x, y, v, a) = (2, 3, 4, 5) for that specific actor).

Stefano

On Wed, Sep 13, 2023 at 4:34 AM Ruud van Gaal <r.vangaal@xxxxxxxxxx> wrote:
Hi Stefano,
 
In my case I'm doing this (with SUMO 1.16):
- moveToXY() with 'keepRoute' set mostly to 0 (once I go off-road I get into trouble if I use any placement bits)
- setSpeed()
- I don't call setAcceleration() (not for a particular reason)
 
In sumo-gui this can give slightly shaking vehicles, but I've attributed that to asynchronicity. Perhaps it's the mismatching acceleration which would make that better.
I didn't ever use setPreviousSpeed(); perhaps this other order helps, although I'm not sure how you define 'get the final wanted result' and if this order would mess that up.
 
Cheers,
Ruud
 

On Wed, Sep 13, 2023 at 12:38 AM Stefano Bonasera via sumo-user <sumo-user@xxxxxxxxxxx> wrote:
Hello,

I am trying to modify 1) position 2) velocity and 3) acceleration of each actor within my simulation to prescribed values that I chose.
Sumo version: 1.16.0 - interfaced with flow
What I currently do is
----------------------------
for loop over all the actors

self.k.kernel_api.vehicle.setPreviousSpeed(
vehID=actor_id,
speed=wanted_velocity - wanted_acceleration*simulation_step)  

self.k.vehicle.moveToXY(
    veh_id=actor_id,
    edge_id=actor.edge,
    lane=actor.lane,
    posX=actor.pos_2d[0],
    posY=actor.pos_2d[1],
    angle=self.k.vehicle.get_angle(actor.actor_id),
    keepRoute=1,
)
 
self.k.kernel_api.vehicle.setAcceleration(
vehID=actor.actor_id,
acceleration=actor.acceleration * 2,
duration=self.sim_step)
 
end for 
 
self.k.simulation.simulation_step()
--------------------------------------
 
With this approach I can successfully set each actor position velocity and accelerations to my values. However it seems the success is really depending on the sequence of steps (for example setting moveToXY prior to setPreviousSpeed results in an unsuccessful final outcome - no error, just not getting what I want).
Therefore, I have the following questions:
0) Is what I am doing 1) feasible and 2) sound, given the above mentioned sequence of commands?
1) I tried to use setSpeed multiple times instead of setPreviousSpeed, but it doesn't work. Can you help me understand why?
2) It seems weird that I have to double my wanted acceleration (see acceleration=actor.acceleration * 2) to get the final wanted result. Why is that happening?
 
Thanks in advance for the help!

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

 
--
Kind regards,
Stefano.

 



Back to the top