TrafficGPT

Installation

TrafficGPT does not require installation, you should just clone the code and run locally.

clone https://github.com/lijlansg/TrafficGPT.git
cd TrafficGPT

The operation of TrafficGPT requires the following software support:

Python
SUMO
PostgreSQL

At the same time, please install the required third-party libraries:

pip install -r requirements

Configuration

LLM Configuration

First, you need to configure OpenAI-Key. Please create a ./config.yaml file in the root directory and add the following content to the file (please modify the content to your own settings):

OPENAI_API_TYPE: 'azure' #'azure'  OR 'openai'
# for 'openai'
OPENAI_KEY: 'sk-xxxxxxxxxxx' # your openai key
# for 'azure'
AZURE_MODEL: 'XXXXX' # your deploment_model_name 
AZURE_API_BASE: https://xxxxxxxx.openai.azure.com/ # your deployment endpoint
AZURE_API_KEY: 'xxxxxx' # your deployment key
AZURE_API_VERSION: '2023-03-15-preview'

Here we recommend using ChatGPT-3.5 to run as LLM. If you want to use your own LLM, please refer to LangChain-Large Language Models to define Your own LLM. In this case, please modify the following sections in ./DataProcessBot.py and ./SimulationProcessBot.py to configure your own LLM.

OPENAI_CONFIG = yaml.load(open('config.yaml'), Loader=yaml.FullLoader)
if OPENAI_CONFIG['OPENAI_API_TYPE'] == 'azure':
    os.environ["OPENAI_API_TYPE"] = OPENAI_CONFIG['OPENAI_API_TYPE']
    os.environ["OPENAI_API_VERSION"] = OPENAI_CONFIG['AZURE_API_VERSION']
    os.environ["OPENAI_API_BASE"] = OPENAI_CONFIG['AZURE_API_BASE']
    os.environ["OPENAI_API_KEY"] = OPENAI_CONFIG['AZURE_API_KEY']
    llm = AzureChatOpenAI(
        deployment_name=OPENAI_CONFIG['AZURE_MODEL'],
        temperature=0,
        max_tokens=1024,
        request_timeout=60
    )
elif OPENAI_CONFIG['OPENAI_API_TYPE'] == 'openai':
    os.environ["OPENAI_API_KEY"] = OPENAI_CONFIG['OPENAI_KEY']
    llm = ChatOpenAI(
        temperature=0,
        model_name='gpt-3.5-turbo-16k-0613',  # or any other model with 8k+ context
        max_tokens=1024,
        request_timeout=60
    )

Fine, now, you can run ./SimulationProcessBot.py.

python ./SimulationProcessBot.py

Database Configuration

Then, in order to run ./DataProcessBot.py, we need to configure the database. Until then, please refer to Github:OpenITS-PG-SUMO Import data into a PostgreSQL database.

Afterwards, your database should contain the following four data tables: topo_centerroad, spatial_ref_sys, zone_roads, the_synthetic_individual_level_trip_dataset. In order to simplify the real-time query operation, two new tables need to be created in the following query sequence: road_level_trip_dataset and road_volume_per_hour.

Create road_level_trip_dataset：


CREATE TABLE road_level_trip_dataset (
trip_id INT,
traveller_id VARCHAR(50),
traveller_type VARCHAR(50),
departure_time TIMESTAMP,
time_slot VARCHAR(50),
o_zone VARCHAR(50),
d_zone VARCHAR(50),
path VARCHAR(50),
duration FLOAT
);

INSERT INTO road_level_trip_dataset SELECT trip_id, traveller_id, traveller_type, departure_time, time_slot, o_zone, d_zone, path::VARCHAR, duration FROM ( WITH split_paths AS ( SELECT trip_id, traveller_id, traveller_type, departure_time, time_slot, o_zone, d_zone, unnest(string_to_array(path, '-')) AS path, duration FROM the_synthetic_individual_level_trip_dataset ) SELECT * FROM split_paths ) AS split_data;

2. Create `road_volume_per_hour`
```sql
CREATE TABLE road_volume_per_hour (
  hour_start TIMESTAMP,
  road VARCHAR(50),
  road_count INT
);

INSERT INTO road_volume_per_hour
SELECT
  DATE_TRUNC('hour', departure_time) AS hour_start,
  path AS road, 
  COUNT(*) AS road_count
FROM road_level_trip_dataset
GROUP BY hour_start, road
ORDER BY hour_start, road;

After all data tables are created, please create a ./dbconfig.yaml file in the root directory and write the following content into the file:

username: your_user_name
password: your_password
host: localhost
port: 5432
db_name: OPENITS

Fine, now, you can run ./DataProcessBot.py.

python ./DataProcessBot.py

Demo

Contact

If you have any questions or suggestions about this project, please send me an Issue and PR, or contact us by email: siyaozhang@buaa.edu.cn

lijlansg / TrafficGPT

readme