DI-smartcross - Decision Intelligence Platform for Traffic Crossing Signal Control

Overview

DI-smartcross

icon

DI-smartcross - Decision Intelligence Platform for Traffic Crossing Signal Control.

DI-smartcross is application platform under OpenDILab

Instruction

DI-smartcross is an open-source traffic crossing signal control platform. DI-smartcross applies several Reinforcement Learning policies training & evaluation for traffic signal control system in provided road nets.

DI-smartcross uses DI-engine, a Reinforcement Learning platform to build RL experiments. DI-smartcross uses SUMO (Simulation of Urban MObility) traffic simulator package to run signal control simulation.

DI-smartcross supports:

  • Single-Agent and Multi-Agent Reinforcement Learning
  • Synthetic and Real roadnet, Arterial and Grid network shape
  • Customizable observation, action and reward types
  • Easily achieve Multi-Environment Parallel, Actor-Learner Asynchronous Parallel when training with DI-engine

Installation

DI-smartcross supports SUMO version >= 1.6.0. Here we show an easy guide of installation with SUMO 1.8.0 on Linux.

Install sumo

  1. install required libraries and dependencies
sudo apt-get install cmake python g++ libxerces-c-dev libfox-1.6-dev libgdal-dev libproj-dev libgl2ps-dev swig
  1. download and unzip the installation package
tar xzf sumo-src-1.8.0.tar.gz
cd sumo-1.8.0
pwd 
  1. compile sumo
mkdir build/cmake-build
cd build/cmake-build
cmake ../..
make -j $(nproc)
  1. environment variables
echo 'export PATH=$HOME/sumo-1.8.0/bin:$PATH
export SUMO_HOME=$HOME/sumo-1.8.0' | tee -a $HOME/.bashrc
source ~/.bashrc
  1. check install
sumo

If success, the following message will be shown in the shell.

Eclipse SUMO sumo Version 1.8.0
  Build features: Linux-3.10.0-957.el7.x86_64 x86_64 GNU 5.3.1 Release Proj GUI SWIG GDAL GL2PS
  Copyright (C) 2001-2020 German Aerospace Center (DLR) and others; https://sumo.dlr.de
  License EPL-2.0: Eclipse Public License Version 2 <https://eclipse.org/legal/epl-v20.html>
  Use --help to get the list of options.

Install DI-smartcross

To install DI-smartcross, simply run pip install in the root folder of this repository. This will automatically insall DI-engine as well.

pip install -e . --user

Quick Start

Run training and evaluation

DI-smartcross supports DQN, Off-policy PPO and Rainbow DQN RL methods with multi-discrete actions for each crossing. A set of default DI-engine configs is provided for each policy. You can check the document of DI-engine to get detail instructions of these configs.

  • train RL policies
usage: sumo_train [-h] -d DING_CFG -e ENV_CFG [-s SEED] [--dynamic-flow]
                  [-cn COLLECT_ENV_NUM] [-en EVALUATE_ENV_NUM]
                  [--exp-name EXP_NAME]

DI-smartcross training script

optional arguments:
  -h, --help            show this help message and exit
  -d DING_CFG, --ding-cfg DING_CFG
                        DI-engine configuration path
  -e ENV_CFG, --env-cfg ENV_CFG
                        sumo environment configuration path
  -s SEED, --seed SEED  random seed for sumo
  --dynamic-flow        use dynamic route flow
  -cn COLLECT_ENV_NUM, --collect-env-num COLLECT_ENV_NUM
                        collector sumo env num for training
  -en EVALUATE_ENV_NUM, --evaluate-env-num EVALUATE_ENV_NUM
                        evaluator sumo env num for training
  --exp-name EXP_NAME   experiment name to save log and ckpt

Example of running DQN in wj3 env with default config.

sumo_train -e smartcross/envs/sumo_arterial_wj3_default_config.yaml -d entry/config/sumo_wj3_dqn_default_config.py
  • evaluate existing policies
usage: sumo_eval [-h] [-d DING_CFG] -e ENV_CFG [-s SEED]
                 [-p {random,fix,dqn,rainbow,ppo}] [--dynamic-flow]
                 [-n ENV_NUM] [--gui] [-c CKPT_PATH]

DI-smartcross training script

optional arguments:
  -h, --help            show this help message and exit
  -d DING_CFG, --ding-cfg DING_CFG
                        DI-engine configuration path
  -e ENV_CFG, --env-cfg ENV_CFG
                        sumo environment configuration path
  -s SEED, --seed SEED  random seed for sumo
  -p {random,fix,dqn,rainbow,ppo}, --policy-type {random,fix,dqn,rainbow,ppo}
                        RL policy type
  --dynamic-flow        use dynamic route flow
  -n ENV_NUM, --env-num ENV_NUM
                        sumo env num for evaluation
  --gui                 open gui for visualize
  -c CKPT_PATH, --ckpt-path CKPT_PATH
                        model ckpt path

Example of running random policy in wj3 env.

sumo_eval -p random -e smartcross/envs/sumo_arterial_wj3_default_config.yaml     

Environments

sumo env configuration

The configuration of sumo env is stored in a config .yaml file. You can take a look at the default config file to see how to modify env settings.

import yaml
from easy_dict import EasyDict
from smartcross.env import SumoEnv

with open('smartcross/envs/sumo_arterial_wj3_default_config.yaml') as f:
    cfg = yaml.safe_load(f)
cfg = EasyDict(cfg)
env = SumoEnv(config=cfg.env)

The env configuration consists of basic definition and observation\action\reward settings. The basic definition includes the cumo config file, episode length and light duration. The obs\action\reward define the detail setting of each contains.

env:
    sumocfg_path: 'arterial_wj3/rl_wj.sumocfg'
    max_episode_steps: 1500
    green_duration: 10
    yellow_duration: 3
    obs:
        ...
    action:
        ...
    reward:
        ...

Observation

We provide several types of observations of a traffic cross. If use_centrolized_obs is set True, the observation of each cross will be concatenated into one vector. The contents of observation can me modified by setting obs_type. The following observation is supported now.

  • phase: One-hot phase vector of current cross signal
  • lane_pos_vec: Lane occupancy in each grid position. The grid num can be set with lane_grid_num
  • traffic_volumn: Traffic volumn of each lane. Vehicle num / lane length * volumn ratio
  • queue_len: Vehicle waiting queue length of each lane. Waiting num / lane length * volumn ratio

Action

Sumo environment supports changing cross signal to target phase. The action space is set to multi-discrete for each cross to reduce action num.

Reward

Reward can be set with reward_type. Reward is calculated cross by cross. If use_centrolized_obs is set True, the reward of each cross will be summed up.

  • queue_len: Vehicle waiting queue num of each lane
  • wait_time: Wait time increment of vehicles in each lane
  • delay_time: Delay time of all vahicles in incomming and outgoing lanes
  • pressure: Pressure of a cross

Contributing

We appreciate all contributions to improve DI-smartcross, both algorithms and system designs.

License

DI-smartcross released under the Apache 2.0 license.

Citation

@misc{smartcross,
    title={{DI-smartcross: OpenDILab} Decision Intelligence platform for Traffic Crossing Signal Control},
    author={DI-smartcross Contributors},
    publisher = {GitHub},
    howpublished = {\url{`https://github.com/opendilab/DI-smartcross`}},
    year={2021},
}
Comments
  • style(hus): update email address

    style(hus): update email address

    Description

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by TuTuHuss 0
  • update and fix typo in docs

    update and fix typo in docs

    Description

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by RobinC94 0
  • update envs, docs and actions

    update envs, docs and actions

    Description

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by RobinC94 0
  • Dev

    Dev

    Description

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by RobinC94 0
  • Merge branch 'main' into dev

    Merge branch 'main' into dev

    Description

    None

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by RobinC94 0
  • update readme

    update readme

    Description

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by RobinC94 0
  • suit for 0.3.0

    suit for 0.3.0

    Description

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by RobinC94 0
  • v0.1.0 update

    v0.1.0 update

    Description

    add cityflow env suit ding 0.3

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by RobinC94 0
  • Dev: Version 0.0.1

    Dev: Version 0.0.1

    Description

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by RobinC94 0
  • Dev: update obs helper, mappo; update configs

    Dev: update obs helper, mappo; update configs

    Description

    update obs helper, mappo; add arterial7; update configs

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by RobinC94 0
  • add different settings for ppo

    add different settings for ppo

    Description

    Related Issue

    TODO

    Check List

    • [ ] merge the latest version source branch/repo, and resolve all the conflicts
    • [ ] pass style check
    • [ ] pass all the tests
    opened by kxzxvbk 0
Releases(v0.1.0)
Owner
OpenDILab
Open sourced Decision Intelligence (DI)
OpenDILab
The NEOSSat is a dual-mission microsatellite designed to detect potentially hazardous Earth-orbit-crossing asteroids and track objects that reside in deep space

The NEOSSat is a dual-mission microsatellite designed to detect potentially hazardous Earth-orbit-crossing asteroids and track objects that reside in deep space

John Salib 2 Jan 30, 2022
School of Artificial Intelligence at the Nanjing University (NJU)School of Artificial Intelligence at the Nanjing University (NJU)

F-Principle This is an exercise problem of the digital signal processing (DSP) course at School of Artificial Intelligence at the Nanjing University (

Thyrix 5 Nov 23, 2022
Flow is a computational framework for deep RL and control experiments for traffic microsimulation.

Flow Flow is a computational framework for deep RL and control experiments for traffic microsimulation. See our website for more information on the ap

null 867 Jan 2, 2023
Numenta Platform for Intelligent Computing is an implementation of Hierarchical Temporal Memory (HTM), a theory of intelligence based strictly on the neuroscience of the neocortex.

NuPIC Numenta Platform for Intelligent Computing The Numenta Platform for Intelligent Computing (NuPIC) is a machine intelligence platform that implem

Numenta 6.3k Dec 30, 2022
Numenta Platform for Intelligent Computing is an implementation of Hierarchical Temporal Memory (HTM), a theory of intelligence based strictly on the neuroscience of the neocortex.

NuPIC Numenta Platform for Intelligent Computing The Numenta Platform for Intelligent Computing (NuPIC) is a machine intelligence platform that implem

Numenta 6.2k Feb 12, 2021
An ML & Correlation platform for transforming disparate data points of interest into usable intelligence.

SSIDprobeCollector An ML & Correlation platform for transforming disparate data points of interest into usable intelligence. At a High level the platf

Bill Reyor 1 Jan 30, 2022
ROS-UGV-Control-Interface - Control interface which can be used in any UGV

ROS-UGV-Control-Interface Cam Closed: Cam Opened:

Ahmet Fatih Akcan 1 Nov 4, 2022
Hand Gesture Volume Control is AIML based project which uses image processing to control the volume of your Computer.

Hand Gesture Volume Control Modules There are basically three modules Handtracking Program Handtracking Module Volume Control Program Handtracking Pro

VITTAL 1 Jan 12, 2022
Data manipulation and transformation for audio signal processing, powered by PyTorch

torchaudio: an audio library for PyTorch The aim of torchaudio is to apply PyTorch to the audio domain. By supporting PyTorch, torchaudio follows the

null 1.9k Dec 28, 2022
git《Pseudo-ISP: Learning Pseudo In-camera Signal Processing Pipeline from A Color Image Denoiser》(2021) GitHub: [fig5]

Pseudo-ISP: Learning Pseudo In-camera Signal Processing Pipeline from A Color Image Denoiser Abstract The success of deep denoisers on real-world colo

Yue Cao 51 Nov 22, 2022
[CVPR2021] Invertible Image Signal Processing

Invertible Image Signal Processing This repository includes official codes for "Invertible Image Signal Processing (CVPR2021)". Figure: Our framework

Yazhou XING 281 Dec 31, 2022
Deep learning (neural network) based remote photoplethysmography: how to extract pulse signal from video using deep learning tools

Deep-rPPG: Camera-based pulse estimation using deep learning tools Deep learning (neural network) based remote photoplethysmography: how to extract pu

Terbe Dániel 138 Dec 17, 2022
The source code of the paper "Understanding Graph Neural Networks from Graph Signal Denoising Perspectives"

GSDN-F and GSDN-EF This repository provides a reference implementation of GSDN-F and GSDN-EF as described in the paper "Understanding Graph Neural Net

Guoji Fu 18 Nov 14, 2022
Code release for the ICML 2021 paper "PixelTransformer: Sample Conditioned Signal Generation".

PixelTransformer Code release for the ICML 2021 paper "PixelTransformer: Sample Conditioned Signal Generation". Project Page Installation Please insta

Shubham Tulsiani 24 Dec 17, 2022
Use MATLAB to simulate the signal and extract features. Use PyTorch to build and train deep network to do spectrum sensing.

Deep-Learning-based-Spectrum-Sensing Use MATLAB to simulate the signal and extract features. Use PyTorch to build and train deep network to do spectru

null 10 Dec 14, 2022
A Simple LSTM-Based Solution for "Heartbeat Signal Classification and Prediction" in Tianchi

LSTM-Time-Series-Prediction A Simple LSTM-Based Solution for "Heartbeat Signal Classification and Prediction" in Tianchi Contest. The Link of the Cont

KevinCHEN 1 Jun 13, 2022
[ICLR 2022] Pretraining Text Encoders with Adversarial Mixture of Training Signal Generators

AMOS This repository contains the scripts for fine-tuning AMOS pretrained models on GLUE and SQuAD 2.0 benchmarks. Paper: Pretraining Text Encoders wi

Microsoft 22 Sep 15, 2022
Repository for Traffic Accident Benchmark for Causality Recognition (ECCV 2020)

Causality In Traffic Accident (Under Construction) Repository for Traffic Accident Benchmark for Causality Recognition (ECCV 2020) Overview Data Prepa

Tackgeun 21 Nov 20, 2022
GNN4Traffic - This is the repository for the collection of Graph Neural Network for Traffic Forecasting

GNN4Traffic - This is the repository for the collection of Graph Neural Network for Traffic Forecasting

null 564 Jan 2, 2023