[sumo-user] SUMO forces maximal deceleration for a vehicle

[Sharon Hornstein]

Hello,
I am running a co-simulation of SUMO (version 1.6.0) and matlab using
traci4matlab.
The simulation includes a merge scenario - two highway lanes that merge into
one.
The geometry includes one roads with two lanes (850 [m] long) , a junction
(zipper, 140 [m] long ) and a single lane (100 [m] long).
The vehicles are advanced using the IDM model which I wrote - I find the
leader and the gap to the leader for each vehicle and calculate the required
acceleration. Then I integrate the acceleration and calculat ethe required
speed and use either setSpeed or slowDown in order to apply the speed for
the vehicle in the next time step.
I use speedMode = 7 ([00111]). MinGap = 2.5[sec]. vehicle length = 5[m].
This works smooth for ~142 [sec] (my simulation runs for 750 [sec]). there
are issues of vehicles that are getting too close to each other while on the
merge, for instance, if two vehicles that are originally on the top and
bottom lane, now are both inside the merge, and they are not aware of each
other, their relative distance can be negative. When this happens, I
deliberately calculate the distance between them and use distance to leader
= 2.5, just that I can continue using the IDM.
At some point, I have a vehicle that gets too close to the other vehicle. I
stop the vehicle using the max deceleration (I use -1.5[m/s^2]) until the
leader is far enough to enable speed update according to the IDM. However,
for several time steps I calculate the required acceleration, then required
speed, I use setSpeed but SUMO ignores the setSpeed command and forces the
max deceleration for the car. The simulation continues but practically this
specific vehicle stops on the merge, thus preventing from the other vehicles
that follow it to proceed.
Apart for this problem, which I can't solve, I also note a parameter "time
gap on lane [sec]" that is found when I right click the vehicle on the
graphical interface. The values of the problematic vehicle keep decreasing
to large negative values (e.g. -140 [sec]). I am trying to understand what
is the meaning of this parameter.
Thank for your help and support,
Sharon

<http://sumo-user-mailing-list.90755.n8.nabble.com/file/t282/car_human013.jpg>
<http://sumo-user-mailing-list.90755.n8.nabble.com/file/t282/car_merge1.jpg>
<http://sumo-user-mailing-list.90755.n8.nabble.com/file/t282/car_parameters.jpg>



--
Sent from: http://sumo-user-mailing-list.90755.n8.nabble.com/
_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

[Jakob Erdmann]

Hello,

with speed mode 7 the bit0 (regard safe speed) is set and therefore sumo can overrule your setSpeed/slowDown command in a situation it considers safety-critical.

You can prevent this by setting a different speed mode or by setting another value for decel and emergencyDecel (note, that more situations will appear dangerous to the model when setting a lower decel value).

the time gap is computed as gap/speed. A negative value means that the leader vehicle has longitudinal overlap or with the ego vehicle (not necessarily lateral overlap). The leader may even be behind ego (but this may be an unexpected edge case related to very long zipper junctions.

regards,

Jakob

Am So., 27. Dez. 2020 um 09:55 Uhr schrieb Sharon Hornstein <[hidden email]>:

Hello,
I am running a co-simulation of SUMO (version 1.6.0) and matlab using
traci4matlab.
The simulation includes a merge scenario - two highway lanes that merge into
one.
The geometry includes one roads with two lanes (850 [m] long) , a junction
(zipper, 140 [m] long ) and a single lane (100 [m] long).
The vehicles are advanced using the IDM model which I wrote - I find the
leader and the gap to the leader for each vehicle and calculate the required
acceleration. Then I integrate the acceleration and calculat ethe required
speed and use either setSpeed or slowDown in order to apply the speed for
the vehicle in the next time step.
I use speedMode = 7 ([00111]). MinGap = 2.5[sec]. vehicle length = 5[m].
This works smooth for ~142 [sec] (my simulation runs for 750 [sec]). there
are issues of vehicles that are getting too close to each other while on the
merge, for instance, if two vehicles that are originally on the top and
bottom lane, now are both inside the merge, and they are not aware of each
other, their relative distance can be negative. When this happens, I
deliberately calculate the distance between them and use distance to leader
= 2.5, just that I can continue using the IDM.
At some point, I have a vehicle that gets too close to the other vehicle. I
stop the vehicle using the max deceleration (I use -1.5[m/s^2]) until the
leader is far enough to enable speed update according to the IDM. However,
for several time steps I calculate the required acceleration, then required
speed, I use setSpeed but SUMO ignores the setSpeed command and forces the
max deceleration for the car. The simulation continues but practically this
specific vehicle stops on the merge, thus preventing from the other vehicles
that follow it to proceed.
Apart for this problem, which I can't solve, I also note a parameter "time
gap on lane [sec]" that is found when I right click the vehicle on the
graphical interface. The values of the problematic vehicle keep decreasing
to large negative values (e.g. -140 [sec]). I am trying to understand what
is the meaning of this parameter.
Thank for your help and support,
Sharon

<http://sumo-user-mailing-list.90755.n8.nabble.com/file/t282/car_human013.jpg>
<http://sumo-user-mailing-list.90755.n8.nabble.com/file/t282/car_merge1.jpg>
<http://sumo-user-mailing-list.90755.n8.nabble.com/file/t282/car_parameters.jpg>



--
Sent from: http://sumo-user-mailing-list.90755.n8.nabble.com/
_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

_______________________________________________
sumo-user mailing list
[hidden email]
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user