satoshirobatofujimoto / pythonpilot

An Open Source Python Framework for Prototyping ADAS and Autonomous Vehicles

PythonPilot

This is an Open Source Python Framework for prototyping ADAS and Autonomous Vehicles. It supports only Lateral Control now.

Table of Contents

• Tested Development Environment
• Getting Started
• Running the demo
• Architecture
• Use Cases
• Acknowledgments
• License

Tested Development Environment

• Processer: Intel Core i9-9900K CPU @ 3.60GHz × 16
• Memory: 16GB
• Graphics: NVIDIA GeForce RTX 2080/PCIe/SSE2
• OS Types: Ubuntu 16.04 LTS 64-bit
• Disk: 240GB

Getting Started

Clone Repository

$ cd ~/
$ git clone https://github.com/YanbaruRobotics/PythonPilot

Prerequisites

• Python 3.5.x

• matplotlib 3.0.2

• numpy 1.16.1

• OpenCV 3.4.5

• scipy 1.2.1

• tensorflow 1.13.0

$ cd ~/PythonPilot/
$ sudo pip3 install -r requirements.txt

Download and uncompress the model

$ cd ~/PythonPilot/
$ wget https://www.dropbox.com/s/i7b6eyzucoxs0fq/models.zip   # 800 MB
$ unzip models.zip
$ rm -r models.zip

Download and uncompress sample log data

$ cd ~/PythonPilot/
$ wget https://www.dropbox.com/s/mo6zo1oo2s46l02/log.zip   # 600 MB
$ unzip log.zip
$ rm -r log.zip

Running the demo

Lunch a vehicle server

$ cd ~/PythonPilot/scripts/
$ bash run_vehicle_main.sh

Lunch a pilot client

Open a new terminal.

$ cd ~/PythonPilot/scripts/
$ bash run_pilot_main.sh

Stop Program

Type 'Ctrl' + 'c' in both terminals.

Architecture

Use Cases

You can see other example videos at here.

Acknowledgments

• The Lane Segmnetation is built upon LaneNet and BDD100K data.
• The Object Detection is built upon realtime_object_detection.

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details

THIS IS ALPHA QUALITY SOFTWARE FOR RESEARCH AND EDUCATION PURPOSES ONLY. YOU ARE RESPONSIBLE FOR USING WITH LOCAL LAWS AND REGULATIONS. NO WARRANTY EXPRESSED OR IMPLIED.