Third-party funded projects

A three year project in the context of the European funding initiative ECSEL (Electronic Components and Systems for European Leadership) with 36 other partners and a total funding of 60 mio. EUR.

In cooperation with Infineon Technologies Dresden we want to add novel image processing approaches into some stages of wafer production. A particular emphasis will be put on merging scalar measurements from other context and image-features. This way, problems are supposed to be discovered ad-hoc or even in advance.

Another field of interest will be the development of low-effort calibration methods.Traditionally, industrial image processing rather favours to carefully tailor the setting of interest rather than making algorithms more flexible. This includes the improvement of lighting, avoiding highlights and silvering, e.g., by shielding reflexions from the ceiling etc. But due to the demand for decreasing setup-times, shorter product cycles or smaller lots, image processing needs to adapt to keep up with demands of Industry 4.0. Just like a human does, algorithms need to cope with reflexions, the lighting available, noise or perspective distortion. 

project website: SemI40
talk to: Frederic Ringsleben

An Indo-German project funded for three years jointly supported by the German BMBF and the Indian DST (Dept. of Science and Technology). It aims for improving civil first-aid in case of flood catastrophes.

In case of a flood-incident first aiders like the THW need an early overview of the affected site for the optimal allocation of personal and resources. To accomplish this we count on the assistance of the local population. Even in remote rural areas of India, the smartphone is ubiquitous. They comprise a number of sensors like GPS, acceleration sensors, radio transceivers for receiving beacons of near-by Wifi stations etc. This the consent of the user, we want to access these readings in order to classify the situation. This might comprise the age-composition of the population, their density or direction of movement. In cooperation with other project partners we also want to detect blocked roads and distinguish flooded from dry areas in order to optimize navigation.

Our particular interest is to derive information relevant for mitigating the catastrophe from all available smartphone readings which one would be expect from analyzing acceleration data for wifi connectivity.

project page: FloodEvac.org
talk to: Maik Benndorf

MiningRoX is part of the RoX-initiative (Robots in SaXony) by the state of Saxony. It has been granted to the TU-Bergbauakademie Freiberg. It is conducted by Prof. Bernhard Jung. The overall objective is to measure and scan the network of tunnels of a mine by using self-propelled wheeled and waterborne robots. They are equipped with stereo-cameras and laser scanners.

In contrast to classical measurements, the true 3D-geometry can be determined and stored.

Within the three years project, HS-Mittweida develops an interactive simulation game of the mine. The game will be used for training critical situations below ground which can not be done in reality for safety reasons. A team-leader defines various incidents like fire, gas leakages or injured miners that need to be found and saved. She is able to see all participating members and their actions.

After the simulation all event can be played back and analyzed by the group jointly.

web-site: MiningRoX
responsible project leader: Thomas Schmieder