Experts of the research group Machine and Computer Vision Applications develop and supply essential components for robotic applications, with a focus on 3D-image processing and reconstruction. Three-dimensional reconstruction of landscapes and objects using cameras and laser scanners has found its way into many applications in recent years. The technology is important for space robotics when unmanned space probes and rovers are sent to explore the surfaces of planets and the moon, and also when surfaces on Earth are mapped for commercial and safety-related reasons.
The methods that are developed and integrated by the research group are ideal for these applications. During the planned 2018 mission of the ExoMars by the European Space Agency ESA, stereo reconstruction of images will be used to help search for life on our neighboring planet, while navigation mechanisms developed at the Institute will serve to plan the daily rover trajectory.
Similar technologies, in the meantime adapted for use in space, are used in tunnel construction, whereby a combination of laser scanners and cameras allows a less than 1 mm resolution level. This technology, thus, provides an important tool for tunnel workers and construction managers to help ensure safety and run cost-effective tests.
One branch of application also deals with surface reconstruction and modeling from 3D sensor data (stereo cameras optionally with a texture projector, 3D time-of-flight and sensors that are similar to those used in game consoles). 3D point clouds that are captured from multiple views (in part at full video rate) can be combined on the basis of their structure and, optionally, their texture, providing a dense network of data for the surface reconstruction. Because of the sensitivity of the sensors to sunlight, these applications are mainly for indoor use.
Machine Vision application areas
Mobility can be defined as the state of being "in motion". Specifically, this branch of research uses machine vision technology to support developments in transport logistics as well as air traffic control.
In the transport logistics sector, robust, real-time sensors are being developed that detect the position of objects in 2D and 3D, as well as new sensors for the static and dynamic detection of the background environment. So that autonomous logistic transporters can fulfill their mission to move goods from point A to B, they must be able to react to changes in the environment. In the simplest scenario, the assumption is made that a personal safety protection device is sufficient for this task.
Generally, a logistics transporter that is located in a rapidly changing environment, for example in large department stores, must be able to analyze the current environment and, where appropriate, plan an alternative transport route. Based on the available reaction time and computing power, the use of alternative and parallel processor architectures (multi-core processors and CUDA platforms to access the graphic card processing power) is required.
In October 2008, the project "Machine Vision Meets Mobility" was supported as a Research Studio Austria program, supporting the investigation of these research topics in collaboration with the Technical University of Graz.
Air traffic controllers face the challenge of processing a continual flood of information. Images and measured values related to the current weather situation are essential. Technology that allows the automation of digital image processing of weather radar and satellite images, the combination of these and the generation of new measurements for improved visibility assessment helps them meet this challenge.