flow
flow copied to clipboard
bump the high of traffic light grid po env
Change the observation space to handle the fact that values go slightly outside of it.
Pull Request Test Coverage Report for Build 5012
- 12 of 12 (100.0%) changed or added relevant lines in 2 files are covered.
- No unchanged relevant lines lost coverage.
- Overall coverage remained the same at 90.356%
| Totals | |
|---|---|
| Change from base Build 5003: | 0.0% |
| Covered Lines: | 8666 |
| Relevant Lines: | 9591 |
💛 - Coveralls
Once that's fixed, LGTM!
Actually I'm curious, what happens when there is no leader, so a vehicle's headway automatically returns min(1000, env.network.scenario_length..)? Your code doesn't actually change this logic, so I'm wondering about how this ever worked. Wouldn't you frequently have a leaderless vehicle returning a really high value?
Also caught one more spot that requires changing, in the update function in vehicle/traci.py around line 222 : self.__vehicles[veh_id]["headway"] = 1e+3 self.__vehicles[veh_id]["follower_headway"] = 1e+3
Oh I missed these reviews! These are all good; thank you!
Actually I'm curious, what happens when there is no leader, so a vehicle's headway automatically returns min(1000, env.network.scenario_length..)? Your code doesn't actually change this logic, so I'm wondering about how this ever worked. Wouldn't you frequently have a leaderless vehicle returning a really high value?
This is actually intentional. We want there to be a default high value indicating that there's no vehicle in front
Actually I'm curious, what happens when there is no leader, so a vehicle's headway automatically returns min(1000, env.network.scenario_length..)? Your code doesn't actually change this logic, so I'm wondering about how this ever worked. Wouldn't you frequently have a leaderless vehicle returning a really high value?
This is actually intentional. We want there to be a default high value indicating that there's no vehicle in front
Ah I see! So a def has_leader(veh_id) function could ostensibly look like if tailway(veh_id) > [some_high_value]: return False ? If it's a developer responsibility to modify their code such that they should be treating ultra-high returns of get_tailway or get_headway as "no leader" or "no tailway" maybe we should advertise that more. But maybe we do? I'm not sure
works like a charm