It was made by a company that owns and rents industrial robotic arms for automating scenic elements and cameras on film shoots so they already had all the expensive gear. It also bears noting that the camera was on an arm as well and this video would not have been possible without carefully matching the camera position to rendered content. From other perspectives in the room, the effect would be much less effective
Here's a blog post from one of the creators. Certain effects are only possible from the single perspective, as @metti indicated. Some of the forced perspective shots would work from a multitude of angles, but would appear differently from various points of view as your mind would attempt to conceptualize it as a three dimensional object.
I suspect the handheld looking shots are also done with a robotic arm that has been programmed using a trainer interface that preserves some of the natural human variation.
Watch the behind the scenes for it that is linked at the end of the video.
