The Technical University of Munich (TUM) receives special recognition for excellent research in the engineering sciences: The European Research Council (ERC) has awarded future funding to three projects from the TUM Department of Informatics and the Department of Electrical and Computer Engineering with ERC Starting Grants. The research projects focus on image processing, data security and cyber-physical systems.
Each year the European Research Council supports future-oriented research projects with substantial ERC Grants, awarded in various categories. Starting Grants are intended for early-career scientists and are endowed with as much as € 1.5 million. The latest awards bring the total number of ERC grants received by TUM to 93.
Prof. Dr. Matthias Nießner (Informatics)
Today's 3D films and computer games are extremely realistic, but have up to now been extremely complicated to produce. It takes graphic artists hours using special design software (CAD programs) to create 3D objects on the computer. Automation of this time-consuming process is the subject of the research being conducted by Prof. Matthias Nießner in his project “Scan2CAD: Learning to Digitize the Real World”. He wants to bring CAD quality to 3D scans, which now often have high levels of noise, no structure or even holes. He intends to develop algorithms which will search databases for a CAD model that resembles the scanned object. Nießner's software will learn how humans design an object in 3D in order to for example show a chair from a static image or a 2D video from all sides. This kind of technology would permanently alter the entertainment industry: Expert skills would no longer be needed to create 3D films. The technology could also be applied in surgical procedures and as a component part in manufacturing facilities for Industry 4.0.
Matthias Nießner is Professor of Visual Computing.
Prof. Dr.-Ing. Antonia Wachter-Zeh (Electrical and Computer Engineering)
Protecting data is enormously important in an increasingly digitalized world – both against unauthorized access and errors during reading and writing. Read/write errors can render major data inventories useless or for example lead to incorrect research results. Prof. Antonia Wachter-Zeh's ERC-funded project inCREASE (“Coding for Security and DNA Storage”) addresses both topics. Advances in computer development, especially with regard to construction of quantum computers, may soon make current encryption methods insecure. Prof. Wachter-Zeh intends to use what are referred to as algebraic codes to develop new encryption methods. Algebraic codes could also help master challenges associated with a storage method that has up to know remained purely experimental: It is possible to store data in DNA and then read it out again. Here however, read/write errors frequently occur. Antonia Wachter-Zeh intends to develop procedures for correcting these errors using the findings achieved with algebraic codes.
Antonia Wachter-Zeh is Professor of Coding for Communications and Data Storage.
Prof. Dr. Majid Zamani (Electrical and Computer Engineering)
Nowadays, there is a diverse variety of so-called cyber-physical systems. All of them contain multiple sensors and actuators are controlled by computer programs. While some are every-day non-critical systems, others, like autonomous vehicles, power networks, or intelligent traffic signals are safety-critical. Simply put, there is no room for error in these systems – a small software bug could have catastrophic consequences. With his project AutoCPS (Automated Synthesis of Cyber-Physical Systems), Prof. Majid Zamani aims at developing a new method for designing cyber-physical systems. It would allow for a mathematical proof that there are no errors in the control code. Formal proofs are already part of design methods in theoretical computer science. These, however, are not suited for the complex problems arising from the physical aspects of cyber-physical systems. Classical control design techniques are also unable to deal with complex requirements needed for cyber-physical systems.
The new approach Zamani plans to develop would benefit manufacturers by removing lengthy and costly testing and validation procedures.
Majid Zamani is Professor of Hybrid Control Systems.