Conversation
AboudyKreidieh
requested review from
cathywu,
eugenevinitsky and
kanaadp
as code owners
Pull Request Test Coverage Report for Build 5763
💛 - Coveralls |
kanaadp
left a comment
kanaadp
left a comment
There was a problem hiding this comment.
Have you checked the performance of manually adding each vehicle vs specifiying an inflow? Seems like it could cause significantly more traci calls?
Otherwise, seems fine. I think this will be really useful.
| accel += np.random.normal(0, self.accel_noise) | ||
| if self.ignore_noise is None: | ||
| # Add noise to the vehicle for all positions in this case. | ||
| accel += np.random.normal(0, self.accel_noise) |
There was a problem hiding this comment.
realized that this should probably be seeded with SUMO seed as well.
| edge=self._inflows[name]["edge"], | ||
| pos=0, | ||
| lane=self._inflows[name]["departLane"], | ||
| speed=depart_speed |
There was a problem hiding this comment.
how does this work if adding a vehicle at the beginning of the lane with the given depart_speed is unsafe? Does it have the same behavior as SUMO inflows (either wait, adapt, or throw away the existing vehicle)?
| from bisect import bisect_left | ||
| import itertools | ||
| from copy import deepcopy | ||
| import random |
There was a problem hiding this comment.
same as above, we should sync the seeds to get reproducibility
| name = names[i + 1] | ||
|
|
||
| # Choose the departure speed. | ||
| depart_speed = self._inflows[name]["departSpeed"] |
There was a problem hiding this comment.
should this also be adaptive?
| else 0) | ||
| ) | ||
|
|
||
| for veh_num in range(num_vehicles): |
There was a problem hiding this comment.
at each time step, we should shuffle the vehicles to be added so they're not added in batches (i.e. 5 humans, then 2 avs)
3 participants
Pull request information
Description
ignore_noiseoption to the controllers to deactivate noise near the start of networksVehicleParamsclass. This gives us more control on how vehicles enter the network.ignore_noisefeature.