Automated lane following of a Waveshare JetRacer with artificial intelligence: Unterschied zwischen den Versionen

Aus HSHL Mechatronik
Zur Navigation springen Zur Suche springen
Zeile 60: Zeile 60:


== Programmierung ==
== Programmierung ==
For the implementation part, a Jupyter notebook has been used, and the code has been
written in Python because it is very user-friendly and it enabled connecting the Windows
PC directly over wifi with the Jetracer.. There is already a library file for the jetracer that
has been developed, and it includes all of the essential functions for the jetracer.
First, using the Etcher program, a prebuilt image that was based on the jetpack 4.5 file
was written to an SD card. This file was retrieved from the official waveshare website.
Next, the jetpack 4.5 file was installed. The NVIDIA JetPack software development kit is
the most comprehensive instrument for the creation of artificial intelligence applications.
All Jetson modules and developer kits may be used with the help of the JetPack software
development kit. Included in the JetPack software development kit are the most recent
Linux Driver Package with the Linux operating system, as well as CUDA-X accelerated
libraries and APIs for Deep Learning, Computer Vision, Accelerated Computing, and
Multimedia. In addition, the software development kit also includes the Linux kernel.
After completing the writing process, the micro SD card was placed into the Jetson nano,
and the device is now prepared to connect to a wifi network using the Jupyter notebook.
The IP address was obtained via the use of the USB connection, and the system may
now be accessed from the personal computer. After successfully logging into the system,
the code that was created in the Jupyter notebooks will be accessible and may also be
changed. So then some Python packages should be installed, and the power mode should
be configured. which can be found in detail on their website  .
First, it was tested using a controller that allows the user to operate the jetracer manually.
This controller is included with the jetracer. Inside may be found a file for a rudimentary
motion and teleportation notebook, which can be used to allow the controller to drive the
jetracer. It is recommended that the controller be linked to the personal computer by way
of the jupyter notebook, which then transmits the signal to the jetracer. The controller
may be callibrated with the use of a library known as ipywidgets. Everything about the
vehicle, including the throttle gain, steering offset, and axes, may be altered using the
Jupyter notebook.


== Komponententest ==
== Komponententest ==

Version vom 5. Januar 2023, 08:41 Uhr

Video 1: JetRacer mit Pytorch programmiert

Autor: Tasawar Siddiquy
Art: Bachelorarbeit
Dauer: 14.09.2022 -
Betreuer: Prof. Schneider


Aufgabenstellung

  1. Mechanischer und elektrischer Aufbau eines JetRacers
  2. Literaturrecherche zu den verfügbaren Programmierumgebungen
  3. Vergleich und Auswahl einer passenden Programmierumgebung
  4. Einarbeitung in die Programmierumgebung
  5. Anlernen des Deep Learning Netzwerkes zur Spurführung
  6. Modul und Systemtests mit dem JetRacer auf der Fahrbahn
  7. Ergebnisdarstellung und Ausblick
  8. Dokumentation des Umgang mit dem JetRacer im HSHL Wiki
  9. Bereitstellung von Demoprogrammen zum einfachen Einstieg
  10. Dokumentation nach wissenschaftlichem Stand

Anforderungen an die wissenschaftliche Arbeit

Getting Started

Nutzen Sie diese Artikel, um sich in das Thema einzuarbeiten:

Nützliche Artikel


Projektplan

Anforderungen

Hardwareanforderungen

Softwareanforderungen

Funktionaler Systementwurf / Technischer Systementwurf

Komponentenspezifikation

  • Specifications of the Jet-Racer

    Specifications of the Jet-Racer

Specifications of the Jet-Racer

Programmierung

For the implementation part, a Jupyter notebook has been used, and the code has been written in Python because it is very user-friendly and it enabled connecting the Windows PC directly over wifi with the Jetracer.. There is already a library file for the jetracer that has been developed, and it includes all of the essential functions for the jetracer. First, using the Etcher program, a prebuilt image that was based on the jetpack 4.5 file was written to an SD card. This file was retrieved from the official waveshare website. Next, the jetpack 4.5 file was installed. The NVIDIA JetPack software development kit is the most comprehensive instrument for the creation of artificial intelligence applications. All Jetson modules and developer kits may be used with the help of the JetPack software development kit. Included in the JetPack software development kit are the most recent Linux Driver Package with the Linux operating system, as well as CUDA-X accelerated libraries and APIs for Deep Learning, Computer Vision, Accelerated Computing, and Multimedia. In addition, the software development kit also includes the Linux kernel. After completing the writing process, the micro SD card was placed into the Jetson nano, and the device is now prepared to connect to a wifi network using the Jupyter notebook. The IP address was obtained via the use of the USB connection, and the system may now be accessed from the personal computer. After successfully logging into the system, the code that was created in the Jupyter notebooks will be accessible and may also be changed. So then some Python packages should be installed, and the power mode should be configured. which can be found in detail on their website . First, it was tested using a controller that allows the user to operate the jetracer manually. This controller is included with the jetracer. Inside may be found a file for a rudimentary motion and teleportation notebook, which can be used to allow the controller to drive the jetracer. It is recommended that the controller be linked to the personal computer by way of the jupyter notebook, which then transmits the signal to the jetracer. The controller may be callibrated with the use of a library known as ipywidgets. Everything about the vehicle, including the throttle gain, steering offset, and axes, may be altered using the Jupyter notebook.

Komponententest

Ergebnis

Zusammenfassung

Lessons Learned

Projektunterlagen

SVN-Repositorium

YouTube Video

Das Video von diesem Projekt finden Sie auf Youtube unter dem Link:

Weblinks

Literatur

→ zurück zum Hauptartikel: Studentische Arbeiten

Abgerufen von „https://wiki.hshl.de/wiki/index.php?title=Automated_lane_following_of_a_Waveshare_JetRacer_with_artificial_intelligence&oldid=98739“