Hi there,
I am trying to randomly place a finite (and specific) set of vehicles in certain parts of a map network in SUMO using Flow along with the Traci simulator but... without using Flow's inflows() objects.
When running the flow.envs.base.reset() to reset the sim with a new randomized configuration, one of the vehicles added is missing.
Let network_object = flow.networks.Network(....) derived from Flow. I can confirm it has the correct ID list and initial state configuration (ID, lane_ID, position, velocity) in its fields. When calling Flow reset(), running self.k.kernel_api.vehicle.getIDList() confirms there are missing vehicles relative to network_object.vehicles.ids. After detecting this discrepancy, Flow errors out.
The error message form is:
File "~/flow_env_with_status.py", line N, in reset
return super().reset()
File "/tmp/ucrl/flow/envs/base.py", line M, in reset
raise FatalFlowError(msg=msg)
flow.utils.exceptions.FatalFlowError:
Not enough vehicles have spawned! Bad start?
Missing vehicles / initial state:
- vehicle_X : ('vehicle_type_X', 'custom_road_name', lane_id, position, velocity)
A hack that seems to help is adding the following code inside of the missing vehicles check before returning the observation vector to the simulation environment:
# check to make sure all vehicles have been spawned
if len(network_object.initial_ids) > len(self.k.kernel_api.vehicle.getIDList()):
missing_vehicles = list(set(self.initial_ids) - set(initial_ids))
for mv in missing_vehicles:
self.k.kernel_api.vehicle.remove(mv) # FIXME: hack
type_id, edge, lane_index, pos, speed = self.initial_state[mv]
self.k.vehicle.add(
veh_id=mv,
type_id=type_id,
edge=edge,
lane=lane_index,
pos=pos,
speed=speed)
self.k.simulation.simulation_step()
# update the information in each kernel to match the current state
self.k.update(reset=True)
I am not sure why the missing vehicle was not added before reaching this part of the code. Is this a problem that others have seen before? Any suggested fixes if we do not want to use Inflows()?
Thanks as always for your time and your help!
Stefano
