How get get values of all the scenarios in paper?

[gaya1thri]

Thank you for the reply, after running the trip info.xml we got this

but in paper 4 scenarios Normal scenarios(normal vehicles, emergency vehicles) and Degraded communication scenarios(normal vehicles, emergency vehicles), are there.How to get results in these cases?

[pangay]

The state changes need to be designed in the file tsc_wrapper.py. Additionally, the insertion of long-tail scenarios should be implemented within the SUMO simulation setup.

[gaya1thri]

Thank you for the reply and your patience, if you don’t mind, could you please eloborate the process and the steps clearly if possible since we have a presentation tomorrow.

[gaya1thri]

Thank you for the reply, after running the trip info.xml we got this but in paper 4 scenarios Normal scenarios(normal vehicles, emergency vehicles) and Degraded communication scenarios(normal vehicles, emergency vehicles), are there.How to get results in these cases?

Thank you for the reply and your patience, if you don’t mind, could you please eloborate the process and the steps clearly if possible since we have a presentation tomorrow.

[pangay]

here’s a clear explanation with examples:
1. Simulating Information Loss: You can randomly simulate information loss by modifying the observation data. For instance, in the code below, a random flag is used to introduce missing information:
flag = random.randint(0, 5) if flag % 5 == 0: # Simulate information loss with a probability observation[0:1, :] = -1 # Mark specific observation as missing infos['information_missing'] = True # Add a missing flag in the info dictionary infos['missing_id'] = 'E0--s' # Specify the ID of the missing information infos['movement_occ']['E0--s'] = '-1' # Set the movement data to missing infos['jam_length_meters']['E0--s'] = '-1' # Set jam length to missing 。

2.Simulating Observation Delays:
Observation delay refers to the agent receiving outdated data instead of the current state. This can be achieved using a queue to store historical observations.

[gaya1thri]

Thank you for the reply, after running the trip info.xml we got this but in paper 4 scenarios Normal scenarios(normal vehicles, emergency vehicles) and Degraded communication scenarios(normal vehicles, emergency vehicles), are there.How to get results in these cases?

(Vtype background,ego,rescue)
Can trying these combinations give us 4 results
Background alone is normal scenario
Background+rescue is emergency scenario
Background+ego is communication degradation
Background+rescue+ego emergency communication degradation, could you please send the code files so that we can run?

[pangay]

The traffic flow files I provided include ambulances. Emergency scenarios are considered long-tail cases, which are overlooked in the design of reward functions for RL models. Regarding communication degradation, the relevant code is already included in the wrapper file. However, communication delays are not currently reflected in the code, and this aspect could be further expanded. For data reading, the built-in functions may not yet explicitly extract and output data for emergency vehicles. I will work on improving this in the future—thank you for pointing it out. Normally, the trip_info.xml file outputs data categorized by vehicle type. You can also try modifying the code to calculate based on vehicle types. Please refer to the following link:
https://github.com/Traffic-Alpha/iLLM-TSC/blob/main/parse_trip_info.py.