Child pages
  • Guided Project WS20_A02 AI Pipeline for Operations Automation at Cologne Bonn Airport

GP WS20_A02: AI Pipeline for Operations Automation at Cologne Bonn Airport

This page sums up the relevant information on GP WS20_A02. It will be frequently updated, so please bookmark this page, if you are a participant or supervisor. 

Current News on the GP

Kickoff for the GP takes place on Wed 16.09. at Cologne Airport. The participants and supervisors know the details.  Agenda for the day at the end of this page. 

The Challenge

Cologne Bonn Airport is the 6th largest freight airport in Europe, with a license for 24/7 flight operation. In the airport operations, several automation and optimization scenarios based in Artificial Intelligence (AI) can be identified.

A possible scenario is pattern detection in the arrival times of freight airplanes. Giving a purely virtual exampe: If the incoming flight from Vienna on Friday night is always late by 60 min in case of rain in Vienna, such a detectable pattern could be used to optimize the resource planning for the cargo ground crews.

Project Definition

In the future, the airport would like to automatize and optimize such scenarios using Machine Learning. This would e.g. mean to automatically scan and analyze all lorry arrivals at the gates, or monitor the flight arrival for patterns. In this project, we will jointly identify the best suitable problem scenario for a proof-of-concept project, and implement a prototype.

As a basis, we will use the CAAI platform (see figure above) developed by the KOARCH project (https://www.koarch.de) in a cooperation between TH Köln (Prof. Dr. Bartz-Beielstein and Prof. Dr. Faeskorn-Woyke) and TH-OWL. This platform follows the pipes-and-filters architecture pattern. There is a working implementation based on Docker and Apache Kafka.

Project Goals

The project goals are twofold:

  1. The project should provide the Cologne Bonn Airport with a prototypical solution for an automated traffic monitoring pipeline. Based on the prototype development, the airport should be able to decide if AI / Machine Learning is worth pursuing further.
  2. Additionally, the CAAI solution should be evaluated for further applications, especially as a possible base for the Data Platform to be built for the Innovation Hub Bergisches RheinLand (IHBR).

Supervision

The project will be jointly supervised by the KOARCH creators and IHBR researchers, with the representative from Cologne Bonn Airport acting as a Product Owner.

Project Mode - 3d/Week Agile Block Project, with Integrated A-Part

We will conduct this project as an agile development, organized in five 2-week sprints with 3d / week working time. The A-part (team supervision and monitoring) will be seamlessly integrated in the sprints. This means that the A-project activities (like e.g. writing a team charter, or performing reflections) will be planned in the sprint backlog, alongside the software development tasks.

This will result in the following schedule:  

Sprint No.

from – to

Development Part

Team Supervision
(“A-part”)

0

14.09. - 25.09.

Kickoff workshop at airport, setting up devenv, getting familiar with technology

Team definition kickoff workshop, introduction to team processes

1

28.09. - 09.10.

Development

Team charter, reflection

2

12.10. - 23.10.

Development

Retrospectives

3

26.10. - 06.11.

Development

Retrospectives

4

09.11. - 20.11.

Development and Documentation

Reflection


afterwards

Finalizing documentation, presentation

Finalizing documentation, presentation, oral exams

Participants of the projects must be prepared to work in a “collocated” fashion. Due to Corona hygiene regulations, this will probably mean “virtual collocation”. It still means, however, that the team must agree on three common working days per week (e.g. Mon – Wed). Each work day starts with a (virtual) standup, and consists of development work in pairs (or at least with continuous presence in a messenger). In similar projects, Discord has work very well for that purpose.

Learning Outcome and Grading

The students will learn to train and run a machine learning algorithm based on a general-purpose platform, i.e. you have the chance to obtain practical knowledge in a type of project that will probably (in one way or another) dominate our IT solution space for years to come.

Formal Learning Outcome

  • As a software developer on Master level, I am able to explore new IT technologies and evaluate their strengths, weaknesses and applicability to a given task, by performing the following steps: 
    1. Capturing the goals for applying the IT technology, by obtaining that information from the appropriate stakeholders,  
    2. Deriving assessment criteria from these goals, 
    3. Defining an small but comprehensive application scenario for a Proof-of-Concept (PoC), 
    4. Capture the business workflows and processes that the scenario is based on, by interviewing business experts (or by other suitable methods), 
    5. Researching the details and background of the technology to be evaluated,
    6. Implementing the PoC-scenario in an agile manner in a team, applying craftsmanship in architecture and coding,  
    7. Describing the PoC results in a compact but comprehensive documentation, so that others can follow my thoughts and conclusions, and can try out my prototype themselves, and
    8. Presenting my PoC project in a lively, interesting presentation, 
  • so that my assessment of the presentation helps others to decide the "if" and "how" questions for that technology.  

Skill requirements in the project 

  • Must:
    • Experience in coding
    • Ability to work in a development team
    • Interest in dealing with complex technical environments
  • Nice to have:
    • Some exposure to machine learning or other AI applications
    • Some experience with hosting environments, especially Docker
    • Apache Kafka experience

Grading

The grading scheme can be found in this PDF (in German):

Meetings

Kickoff 16.09. at Cologne Bonn Airport

Agenda (in German)

TimeTopicAgenda Items
9:00Intro
  • David: Flughafen und Use Case vorstellen (ca. 15 min)
  • KOARCH vorstellen (ca. 30 min)
  • Innovation Hub (als potentieller Anwender von KOARCH) vorstellen (Julian, ca. 15 min)
  • Benotung und Struktur des GP (Stefan, Eberhard - 15 - 30 min)
10:30Pause
10:45Use Case / Szenario
  • Begehung im Flughafen
  • Präsentation der verfügbaren Daten (David)
12:30Mittag
13:15Organisation
  • technische Orga-Fragen (ca. 30 min)
    • Randbedingungen
    • Tools
    • Speicherung von Daten und Source Code
  • Projektorganisation
    • agile Struktur
    • Einbettung der GP-A-Anteile
~ 15:00 - 16:00Ende