Künstliche Intelligenz / Artificial Intelligence (Praktikum)

Dozent(in): Prof. Dr. Martin E. Müller
Termin: Mo. 14:00 - 18:00
Gebäude/Raum: 101

Inhalt der Lehrveranstaltung:

Praktikum "Künstliche Intelligenz "

  • Wie im letzten Jahr vertiefen wir die Grundlagen aus der Vorlesung durch prkatische Arbeit an Lego-Robotern.
  • Das Praktikum und die Vorlesung können nur gemeinsam besucht werden - für beides zusammen gibt es 8 LP.
  • Registrierung siehe Praktikumsseite.

Lab Session "Artificial Intelligence"
The aim of this lab session is to gain a deeper understanding of AI problems and techniques as presented in the lecture by trying to put them into practise. This lab session takes 4 hours a week being present at the lab, but you will be needing and wanting to spend much more time on the project.

Like in the last year's course, we will be working with Lego Mindstorm kits. To speed up the process of putting AI onto the bots, we skip the robot design and you will be provided with a kit and a predefined robot design (including sensors etc.). The tasks the root will have to solve are as follows:

  1. Navigate through a maze and avoid to bump into walls.
  2. Navigate through a maze without bumping into other moving objects.
  3. Some kind of robot sumo One-on-one or maybe even in dynamic groups. (See Simon Parson's AI/Robotics teaching page).
The robots will be equipped with various stock sensors, some homebrewn stuff, and selfmade bridges that allow the robots to communicate with a PC via WLAN (I.e. the Lego RCX unit will act as the cerebellum receiving primitive commands from the AI-"cerebrum" running on the PC). Additionally we will use a top-down camera view on the arena that allows the robots to check their actual spatial orientation against their internal model and which in a second step shall provide the robots with a fake first-person-view on the scene.

Programming Tasks
There are two different kinds of programming tasks ("exams") to be solved during the lab sessions:

  • Intermediate Tasks ("bi-weekly assignments")
    1. Symbolic world representation
    2. Graph search:
      Routing and navigation
    3. Planning:
      Oneplayer and team tactics
    4. Learning:
      Reinforcement Learning for motor control and strategy learning.
  • Final Tasks ("final exam")
    1. Maze navigation
    2. Obstacle Avoidance
    3. Robot battle field
    4. Presentation & Paper


weitere Informationen zu der Lehrveranstaltung:

empfohlenes Studiensemester der Lehrveranstaltung: ab dem 6. Semester
Fachrichtung Lehrveranstaltung: Multimedia-Methoden
Dauer der Lehrveranstaltung: 4 SWS
Typ der Lehrveranstaltung: P - Praktikum
Leistungspunkte: 8
Semester: jedes WS