---
type: slide
slideOptions:
theme: white
center: false
---

<style>
.reveal {
font-size: 2.5rem;
}
</style>

<p align="center">
<img src="https://raw.githubusercontent.com/dora-rs/dora/main/docs/src/logo.svg" width="400">
</p>
Low latency, distributed, and composable data flow framework.

[github.com/dora-rs/dora](https://github.com/dora-rs/dora/)

<br><br>
Xavier Tao, [e-mail]
Nov 17, 2022

---

## Objectif du projet

> If we were to rewrite ROS today, what would it look like?

- Introduire un nouveau framework de robotique avec:

- Minimiser la latence de communication
- Maximiser la composabilité
- Améliorer l'ergonomie par rapport ROS/ROS2

---

## COMMUNICATION

- Utilisation de `shared memories` pour les communications inter-process.

> tcp: Average duration: 44.391 us
> shared memory: Average duration: 0.238 us
> Message size: 128 bytes
> See: https://stackoverflow.com/questions/2854004/ipc-speed-and-compare

- Utilisation de `zenoh` pour les communications en système distribué

- Minimisation du nombre de copies des données

---

## COMPOSABILITE

- Utilisable sur Python, Rust, C, C++
- Réutilisation d'operateur existant similaire à dockerhub.

---

> Exemple de demo Python: https://github.com/mikel-brostrom/Yolov5_StrongSORT_OSNet/blob/master/track.py

---

communication:
  iceoryx:
    app_name_prefix: dora-iceoryx-example

nodes:
  - id: webcam
    custom:
      run: ../../physicals/webcam.py
      inputs:
        timer: dora/timer/millis/100
      outputs:
        - image

  - id: yolov5
    operator: 
      outputs:
        - bbox
      inputs:
        image: webcam/image
      python: ../../operators/yolov5_op.py

  - id: strong_sort
    operator: 
      outputs:
        - obstacles_id
      inputs:
        image: webcam/image
        obstacles_bbox: yolov5/bbox
      python: ../../operators/strong_sort_op.py

  - id: plot
    operator:
      python: ../../physicals/plot.py
      inputs:
        image: webcam/image
        obstacles_bbox: yolov5/bbox
        traffic_sign_bbox: traffic_sign/bbox
        lanes: yolop/lanes
        drivable_area: yolop/drivable_area
        obstacles_id: strong_sort/obstacles_id

---

## Extensibilité

### Opentelemetry Tracing

Protocol de tracing en système distribué dans plusieurs langages.


> Opentelemtry: https://opentelemetry.io/docs/concepts/signals/traces/
> Jaeger: https://www.jaegertracing.io/docs/1.21/opentelemetry/

---

### Opentelemetry Metrics


---

### DB supportant opentelemetry

- Datadog
- Prometheus
- InfluxDB
- ...

> Future Cloud Native integration: AWS S3, IAM, HashiCorp Vault, Keycloack, Docker, Kubernetes, ...

---

## Ergonomy

Simple installation

wget https://github.com/.../dora-<version>-x86_64-Linux.zip
unzip dora-<version>-x86_64-Linux.zip
PATH=$PATH:$(pwd):$(pwd)/iceoryx
dora --help

Readable description

communication:
  zenoh:
    prefix: /abc_project

nodes:
  - id: webcam
    operator:
      python: https://raw.githubusercontent.com/dora-rs/dora-drives/main/operators/webcam.py
      inputs:
        tick: dora/timer/millis/100
      outputs:
        - image
  - id: plot_window
    operator:
      python: https://raw.githubusercontent.com/dora-rs/dora-drives/main/physicals/plot.py
      inputs:
        image: op_1/image

Command line program

dora up 
dora start dataflow.yml

---

### Questions?

---

### Comment pouvons nous avancer ensemble?

- Vous aider à accomplir des projets ambitieux.
- Développer le Campus de CentraleSupelec, l'infrastructure existente
- Créer des partenariats internationaux avec CentraleSupelec. Huawei, c'est 22 Milliard $ de Budget R&D. En comparaison Safran, c'est 1,2 Milliard $ de budget R&D.

---

### Projet d'une voiture autonome


---

### Essai IRL

https://www.instagram.com/p/CkquyYJJb4y/