From what I can see it's just a regular parallel SCARA, the only difference is one of the driving links has been parallelized to prevent rotation at the effector. This is usually unnecessary, you can deal with the hotend rotating a little along the z axis, but can be useful if you have multiple hotends in order to turn things into a fixed offset rather than yet another trig function to bog down the processor (to deal with the rotation of the effector).