The video has a English dubbed audio track. The actual tour starts around 8:00.

Just one sturgeon? It is easy to get lost in that labyrinth, make sure you biring a torch. And yes, we find Ibbenbüren a place to visit every year

If I was (still) working at a University this would be a great starting point for a paper. Now, being an independent development, I need to find a way to turn this into new work...

The rest (cohesion factor, alignment factor, separation factor and separation distance) are all weights in the network and can therefore be trained...

- cohesion is just the sum of the relative positions of the neighbours;
- alignment is the average of the relative velocities;
- separation is the sum of relative position with the magnitude = 1/v where v is the original magnitude.
(The last one is still a bit too complex: WIP)

Solution is to add a Collider to the avatar so that the robot can detect it. This is actually (surprisingly) difficult to achieve using the GLTF model format.

An interesting fact is that the Unity instance on the left does not have any knowlegde about the Humanoid in its project assets. This contrasts with other networking solutions where every instance needs to have the same Unity assets.

Quest Pro, although I hate how fast Meta forgot about it, it is the only one where I can keep my glasses on.

I would _love_ to make a dive in the NBF with Columbus, but they didn't let me...