Campus FreeCity

HUMAN-MACHINE INTERACTION

The CityBot is designed to offer a human-machine interface (HMI) that does justice to people's means and habits of communication. Central University Administration is the avatar head of the CityBot, combined with other components such as visual displays. Interaction with the system must be designed to be low-threshold and connect to the user's experiences with pass transport systems in public spaces. To this end, a multimodal approach is being realised that is easy to understand and should function robustly. Multimodality stands for the combined use of several input/output methods, in the project speech, gestures and visual elements.

User requirements and experiences are systematically incorporated via a human-centred design process, implemented via a multi-step, iterative development including usability evaluation. User groups (customers, passers-by, staff) and use cases are analysed, a scenario catalogue is created, the interaction concept is developed iteratively with interactive applications in virtual reality and finally evaluated in the real-life laboratory phase.

The results are fed back into the technical design of the system. At the end of the project, a systematically evaluated interaction concept for the CityBot avatar will be available that precisely covers the use cases, is technically feasible and meets the requirements of practical use.

AREA OF IT SECURITY

With its modular and flexible architecture, the CityBot differs greatly from previous, static automotive architectures. This modularity brings with it new challenges and risks, particularly in the area of IT security, as interfaces and processes that were previously assumed to be static are suddenly openly accessible to external devices.
In order to identify, understand and address the resulting risks at an early stage, Prof Dr Michael Zohner's working group is analysing security and developing a security architecture for the CityBot. The security architecture will pass through proven methods in the field of automotive security as well as new mechanisms developed specifically for the CityBot, which take into account the dynamic nature of the CityBot infrastructure. The mechanisms developed are to be integrated into the CityBot and evaluated there in order to provide impetus for a later implementation that complies with examination requirements.

AREA OF ENERGY-EFFICIENT KI

Artificial intelligence (AI) methods provide key technologies for the realisation of autonomous vehicles such as the CityBot. Many of the AI tasks must be performed in real time and thus in the vehicle (edge computing). This is very computationally intensive and requires high energy consumption on CPUs/GPUs, which is problematic in terms of battery life and cooling. Field Programmable Gate Arrays (FPGAs) are known for realising AI tasks in a very energy-efficient and high-performance manner. Compared to GPUs, FPGAs only require around a tenth of the energy that a GPU implementation needs for the same throughput. In addition, FPGAs allow much shorter latency times, which is important for real-time processing.

However, the development of efficient FPGA architectures is time-consuming and requires expert knowledge from AI and chip design. The aim is therefore to develop domain-specific design software that automates large parts of the architecture design and allows realisation in the CityBot system without the need for extensive expert knowledge. The special features of the FPGA should already be taken into account when training the neural network in order to achieve high energy efficiency with full accuracy.

AREA OF SUPPLY CHAIN INTEGRATION

In addition to the numerous applications of the CityBot in mobility, the autonomous vehicle will also be able to handle comprehensive logistics tasks. Where large lorries previously provided large parts of the supply chain in city centres, the new vehicle represents a low-emission and reliably plannable logistics alternative. The handling of goods at the edge of the SmartCity and the process modelling of logistics chains represent exciting challenges in this vision of the future.

These logistics aspects in the future SmartCity are being investigated by members of the Department of Business. In collaboration with the Bundeswehr University in Munich, algorithms are being developed to plan and execute the CityBot's supply chain processes at a strategic, tactical and operational level.

WEBSITE

https://www.campusfreecity.de