The Future of Education

The development of Artificial Intelligence (AI) has been receiving increasing attention and is accompanied by efforts to adapt teaching methods accordingly. Supported by the German Federal Ministry of Education and Research (BMBF), we aim to integrate fundamental AI concepts into various faculties at our university. A particular focus is placed on hands-on teaching approaches, especially through practical examples for entry-level and advanced students, as well as for school education. Our work focuses on the integration of AI with robotics, aiming at the physical manipulation of objects. Building on previous work, we report significant progress in this area. We implement a three-stage teaching concept characterized by increasing levels of proficiency: student level, university entry level, and advanced university level. In this presentation, we briefly report on our overall experiences and then focus on progress and future plans at the most advanced level. The course introduces students to the integration of both an open-source and a commercial AI system into the control system of a robot. Our experience shows that the implementation phase has the greatest educational impact, as it provides low-level access to the code required to interconnect the AI network with the robot’s electronic interface. This revealed an unexpectedly high level of complexity and highlights the true demands placed on engineers. However, our initial approach also revealed several drawbacks. Incorporating the open-source model proved to be more complex than anticipated, while the commercial AI system was far more accessible. At the same time, the code segments generated by the students are embedded within a larger software environment, which makes understanding unnecessarily difficult, as the complex user interface obscures parts of the underlying functionality. Moreover, the significant cost of the commercial system in both hardware and software renders it unsuitable for replication on multiple computers. To address these issues, we developed a self-programmed AI and AI-interface system using an open-access AI framework. In our limited exercise scenario, it recently achieved the same robot control functionality as the far more powerful—but largely idle—commercial system. Our solution can be replicated on minimal IT resources, is cost-free, and provides full access to all code levels, enabling students to gain deep insight into the system. We report on the implementation, performance, and the time and skill demands placed on students when developing such a system. Finally, we discuss our didactic experiences and outline plans to adapt our course to incorporate the new AI model template.

Keywords: Artificial Intelligence, robotics, algorithms, coding