After my master's thesis, the Institute of Machine Tools and Production Technology (IWF) got in touch – they had seen my work on Co-SpACE#. The IWF was building a learning factory that, alongside physical machines, was to offer a learning platform and mixed reality support. I was part of the team for the ILIAS-based learning platform and at the same time solely responsible for the MR platform.

The platform was meant for students, few of whom own an MR headset – so it needed good fallbacks and had to be easy to deploy. Since in the long run not only students but also employees of external companies were to learn here, licensing with engines like Unity or Unreal was difficult. For all these reasons I chose Godot – a decision that quickly convinced me on other grounds too.

A web export makes the platform immediately reachable for every student. Assets live in a CMS I wrote myself and are loaded only on demand, which lets teaching staff design their own scenes without writing a line of code. The platform is most impressive in an MR headset, but runs just as well on phone and desktop. A self-written multiplayer layer enables simple interaction between several people, and a robust step-by-step system presents instructions and interactive information.

The system is entirely self-contained and needs no external services – the lesson from Co-SpACE#. Instead of spatial anchoring via Azure I developed finger-based anchoring: you place your fingers on two defined real-world points at the start, and the system works out where it is and where the virtual elements belong. In my tests the accuracy was, surprisingly, under a centimetre – which makes the method quite competitive.

Due to various circumstances at the IWF I couldn't develop it beyond the proof of concept. In its simple, robust state it is surprisingly usable nonetheless.