Enabling Human-Machine Interaction in Real Time

A drink at the bar - "shaken, not stirred" - prepared exactly the way every super agent likes it. But in this bar, the bartender gets help from a cobot, a collaborative robot. The cobot can be seen live at the Center for Tactile Internet with Human-in-the-Loop (CeTI) at the Dresden University of Technology (TU Dresden). It works together with people - in close proximity, without safety devices, as its sensors detect obstacles and automatically switch it off in an emergency.

Of course, the whole thing is more of a test scenario - in future, the cobot will feel particularly at home in places where automated production takes place. Here it can take over assistance functions. It only needs data transmission of 5G or faster. This is because the latency time for interaction between humans and machines must be less than one millisecond, which is the only way to ensure that the complex movements made by a bartender (or assembly worker) are fluid. Interdisciplinarity plays a major role in this CeTI project, where not only computer scientists are working on the next stage of the Internet, but also communications engineers, mechanical engineers, psychologists and philosophers.

The vision of CeTI, a Cluster of Excellence at the TU Dresden, is to enable people to interact with cyber-physical systems in the real or virtual world via intelligent communication networks and systems in near real time.

Such advances go far beyond the current state-of-the-art approaches in computer science and engineering: intelligent communication networks and adaptive CPS for quasi-real-time cooperation with humans require online mutual learning mechanisms, which are a key challenge.

To address these challenges, CeTI conducts unique interdisciplinary research and addresses important open research topics in key areas of human-machine control complexity, sensor and actuator technologies, software and hardware designs, and communication networks as a basis for several novel use cases in the fields of medicine, industry and the Internet of Skills.

In this context, CeTI supports research on cross-cutting AI topics that are relevant to a larger number of related research areas and fosters collaboration with other AI initiatives. The scientists ask themselves questions such as:

• How can sensory modalities (vision, speech, action, touch, smell) be combined?
• How can AI be optimized for resource-constrained distributed environments?
• How can reliability and confidence in AI-based decisions be ensured?
• How can AI models from different paradigms be integrated?