direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

Competences

Lupe

The focus of our research is the development of methods and technologies for the coupling of industrial and process automation with secure global communication and private clouds. Increasingly, real-time applications are being addressed, taking into account latency and security requirements. Our core competencies are amongst others in industrial protocols, industrial cloud applications, safety requirements and standards of process automation as well as key paradigms of the Industrial Internet and applications in M2M and IoT. In the following, there are listed examples of communication networks and communication middleware, as well as a broad presentation of ​​industrial application fields.

Communication Networks
Communikation Middleware
Mobile radio LTE und 5G
OPC UA + TSN (Time Sensitive Networking)
Industrial BUS-Systems 
Robot Operating System (ROS) und ROS 2
...
...

Fields of Application

Lupe

Cloud Robotics

In the field of robotics, a paradigm shift from monolithic stand-alone solutions towards distributed cloud-based solutions for control and administration can be seen recently. Simply spoken, , cloud infrastructures provide access to seemingly unlimited computing and storage resources that can be used in a location-flexible manner. Accordingly, more complex staging architectures and more efficient control algorithms can be realized. This favors the increasing use of artificial intelligence methods. Furthermore, the modularization and service-oriented design of control architectures results in more flexible approaches for integrating software components and managing their lifecycle. Platform offerings have the potential to serve as universal software marketplaces for robots. Skills and knowledge of a particular robot can therefore be shared and commercialized with other robots arounf the world with little effort. In addition to these advantages, there are also challenges in the areas of scaling and virtualization, connectivity, networks as well as security and privacy.

Flexible industrial system through Artificial Intelligence

Machine learning methods are increasingly being used in industry in the form of intelligent systems and enabling flexible and adaptable processes and applications.

In recent years, deep learning techniques in particular have revolutionized many applications. The performance of many classical approaches could be significantly exceeded, e.g. in image classification. The basis for the powerful algorithms are huge amounts of data and the processing power of GPUs.

Since direct use of GPUs in machines and robots is usually difficult or impossible (e.g. for mobile robot systems), deep-learning applications in industry are mostly realized as distributed architectures. For example, image data is transferred from a robot-integrated camera to a factory cloud, which performs image-based evaluations based on deep learning and then sends the result back to the robot. There are further approaches to more efficient handling of data, for example by using previous knowledge in the field of transfer learning and the targeted learning of similarities / differences through metric learning.

Industrial Augmented Reality

Applications of the Augmented Reality (AR) enable the spatial visualization of virtual information (e.g. virtual objects, text, etc.) in real environments. Characteristic for the AR is a spatial relationship between virtual and real objects. Thus, we could provide simulation for  robotics and process automation within real environments. Further industrial applications are the training of employees and the process-integrated support of humans through spatial information.

AR display or display media can be implemented as either optical see-through or video see-through, depending on whether the virtual objects are displayed on a transparent display or in a camera image. There are also experimental technologies for retinal displays. The visualization can be carried out both via static devices (e.g. fixed display and fixed camera) and via mobile devices (e.g. tablet PC or glasses with integrated or external camera). In order to realize a perspectively correct representation of the objects within (mobile) AR allications, a (continuous) localization (also registration or continuous: tracking) of the AR display medium or the associated camera in the environment is needed. This localization is mostly done using 2D or 3D camera images. Methodically, this can be achieved by tracking artificial markers or by tracking inherent environmental features using Simultaneous Localization and Mapping (SLAM).

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe

Auxiliary Functions

Contact

Prof. Dr.-Ing. Jens Lambrecht
+49 (0)30 835358412