TUM: Science Hackathon 2021 – Season 2

Are you ready for one of the most exciting tech events of the year?

The next TUM: Science Hackathon takes place this autumn!

  • Kick-Off: November 18, 2021
  • Main event: November 26-28, 2021

The application process is open until November 5th, 2021!

Register individually or as a team and have the chance to win one of the prizes:
1st Winner: 1000€

2nd Winner: 600€

3rd Winner: 400€

Click here to register

Students and Ph.D.-students from diverse fields of study from all universities of Munich are invited to participate in the #ScienceHack. They display their creativity, passion and problem-solving skills towards solving challenges provided by the industrial partners and exceptional chairs of TUM.

For the first time, students and Ph.D. students of the EuroTeQ-partner universities are invited to participate!


What is the topic of the Science Hack 2021-2?

This Science Hack focuses on “Human-Centered Technology”. The momentum of developing new technologies has been unbroken since the invention of the first computers and the resulting possibilities of digitalization. However, this impressive shift is only so helpful if people are at the very center of it. In other words, technological breakthroughs should serve people, improve their lives, and secure their future. Traditional technologies experience disruptions, are replaced by new ones, are subject to new requirements, expectations, and target groups. In the constant search for innovations by companies and scientific pioneers, however, numerous developments often do not follow human benefit, but merely trends.

The industrial partners of the 4th Science Hack are:

The challenge: Envision The Bike Experience of Tomorrow

Find out more about the challenge: Altair Challenge

The challenge is to develop a route planner that plans routes with the lowest possible CO2 demand, while also keeping an eye on the movement destinations and the different modes of transport (e.g. car, public transport, shared drive, bike share,...). In addition to the selection of the means of transport, the planner also takes into account other parameters such as car class, performance/range, and fuel/ride price.

Furthermore, an optimization of time, distance, and CO2 should be considered. The goal is to develop a website or an app that is easy to handle, understandable, and easy to use.


  • Algorithm development and optimization procedures
  • Programming language (C ++, C #, Javascript or similar)
  • Web or app development
  • UX design • Graph-based methods




Health awareness in society is increasing rapidly and with it the consumption of healthy food grows. The ecological footprint of food production must be as small as possible to ensure sustainable food production. At the same time, there is a growing interest in the "story" behind the food we eat, requiring more transparency in the life cycle of food.

One approach to achieve this could be more decentralized agriculture, especially for healthy plant-based superfoods. Urban agriculture on roof tops, in backyards etc., considering short supply routes, circular production and smart resource use could help solve such an approach.



  • How could a concept look like?
  • How could a smart garden be realized?
  • Who should/will run smart gardens?
  • Who wants to check the origin of the food?
  • What could a mobile smart green bio app look like?

"Is my chat bot empathic enough?"

The challenge: Humanizing emotionally challenging dialogues with machines

Find more about the challenge: Pixida Challenge

Specific skills (such as JavaScript, R, ...) are not required for the challenge


Every day ProSiebenSat.1 produces thousands of bytes of valuable media content. The idea is to use the possibilities of artificial intelligence and cloud technologies, to improve economic potentials in the field of classic TV broadcasting and to improve the utilization of multimedia content across different platforms and content types.The goal is to provide the right content based on business needs, delivered in an automated system that is able to be queried by consuming applications. It is given that ProSiebenSat.1 media sources provide metadata and tags already. However, please feel free to consider this part in your planning as well.


Create an architecture where content metadata can be stored and be further used by consuming systems.

Your goal is to provide the right content based on these requests.

A typical business request for content could be:

“ We need a diverse group of happy people in stunning scenery, at best outdoors, which are enjoying the moment. Content must be FSK12 friendly.”

Find more about the challenge: ProSiebenSat.1 Media SE

The challenge: IoT Food Delivery Monitoring

Find more about the challenge: Sigfox Germany GmbH Challenge


The cultural change in a company is, among other things, also an essential basis for the innovation and sustainability of a company. One way to promote this is through the use of new methods. Gamification is now being used successfully in a wide variety of contexts. Within the company, however, gamification approaches have so far played a subordinate role, and they have not been used at SWM so far. SWM would like to change this, although the area of application is still open. The question arises in which internal company context this makes sense and what kind of implementation comes into question.


Implement a gamification approach for SWM employees; in which situations could a gamification approach be usefully applied in the company; implement a corresponding application.

Additional challenges are provided by three chairs/ professorships of TUM.

The challenge: Deep Learning for Object Detection in The Context of Smart Agriculture

Find more about the challenge: Professorship of Precision Agriculture Challenge

The challenge: Enabling citizen participation in transportation infrastructure decisions

Find more about the challenge: Chair for Traffic Engineering and Control Challenge


Deepfakes have permeated many aspects of our daily lives. Any news media organization will be confronted with deep fakes, and credibility and trustworthiness are at stake. We need methods for analyzing audio and video content to make sure they have not been produced or altered with methods commonly used for creating deep fakes


"Setting up an easy-to-use toolbox for manual forensic analysis of deepfakes (audio and video)" with the following characteristics:

  • made for journalists who have some technical background
  • all tools based on machine learning should be trained and ready to use
  • the toolbox should include a number of examples and there should be an indication as to how reliable the results are, i.e. how likely it is to find false positives or false negatives
  • the toolbox should be installed on a dedicated physical system, but in case a lot of processing power is needed, it can be installed on any cloud service, preferably as a container image  
  • operating systems: Windows 10, Ubuntu, or running on a cloud service
  • documentation of the forensic process


  • IT: software development, data analysis, video and audio formats, deployment of container images
  • Basic knowledge about deepfakes and their production process