autocore-ai / sdv

AutoCore SDV Project

For business and partnership, please visit our website: www.autocore.ai.

Table of Contents

1. Overview
2. Quick Start Guide

Overview

SDV (Software Defined Vehicle) project is a cooperative project between Autocore and Futurewei, which aims at providing the technology-consulting services by means of this 100 percent open source software stack with reference design for SDVs. Cloud-Edge service is also integrated as a part of this project to extend the V2X capabilities.

The SDV software stack is based on open source Autoware/ROS2/DDS and Zenoh, where DDS/Zenoh act as the E2E Vehicle-Edge-Cloud middleware layer for the SDV platform. Thru the integration with Futurewei’ s open source KubeEdge project, the SDV platform will leverage Cloud Native Ecosystem tools to provide management, monitoring and software LCM (Life-Cycle-Management) functions.

The system architecture of SDV platform project is shown in the figure below:

The software modules in host container runtime are as follows:

And the PCU Container runtime acts as the Domain controller in vehicle with the following software modules:

Quick Start Guide

Hardware requirement

• Host:

  • Host PC:
    • CPU: x86_64 4 Core or above
    • RAM: 8G+
    • Disk: 30G+ free space
    • OS: Ubuntu 18.04+
    • cable Ethernet

• For each Worker (default sets 2 workers in cluster):

  • AutoCore PCU

    • Flash images from Release for k8s install

    • cable Ethernet

      OR

  • Worker PC (Needs when you don't have AutoCore PCU):

    • CPU: x86_64 8 Core or above
    • RAM: 8G+
    • Disk: 30G+ free space
    • OS: Ubuntu 18.04+
    • cable Ethernet

Environment setup

1. Host PC:

• $ source Utils/setup/k8s/control-plane/setup.bash

  Concole will output the information like kubeadm join xxx, please use this information for clients to join in later steps.

1. AutoCore PCU OR Worker PC:

• $ source Utils/setup/k8s/worker/setup.bash

• $ sudo kubeadm join xxx

  Please find the xxx in the concole output of Host PC as described in step 1.

1. Back to Host PC

• For each worker node, label them:

  $ kubectl label node <worker_node_name> app=autoware

Deploy workloads with config file

Default configs 2 workers in cluster, and if you have two works hardware, just run the default command:

$ kubectl apply -f https://raw.githubusercontent.com/autocore-ai/SDV/develop/sdv_demo.yaml

And if you have other custom deployment, Just download sdv_demo.yaml and fix the replicas: 2 to your worker count N:

spec:
- replicas: 2
+ replicas: N
  selector:

then use this config file:

$ kubectl apply -f sdv_demo.yaml

Use CloudViewer to display scene and send commands.

Open CloudViewer with Chromium based browser

• Config IP Address to Host PC IP (defaults 127.0.0.1)

• Move viewer with key WSAD and mouse midde and right key.

• Click on vehicle and traffic light to inspect and send commands.

Video tutorial

sdv's Issues

Zenoh on MCU porting

V2X Solution design markdown

V2X solution design and realization (in Simulator)

OS and Middleware baseline function description markdown.

OS and Middleware baseline development

Zenoh integration

Zenoh on MCU porting and Intergration markdown

CloudViewer Document