I think the Abacus Drive by SRI is a really interesting new type of gearbox for robotics and force control. It’s cheaper, simpler to produce & more efficient than a harmonic drive & still has near zero backlash. Also, unlike the harmonic drive, it can be back driven to allow force sensing through the motor. Unfortunately it looks like it’s going to be wrapped up in patents (but there have been a couple of attempts at 3d printing one 
).
When the article was first published last year I read it with great enthusiasm and the feeling that finally the Harmonic Drive AG stranglehold on robotics would be over, only to get to the end of the article and be utterly crushed to find that they have given the rights to Harmonic Drive AG… More sand in the gears of industry.
The only info I can see pertaining to the release of this Cycloidal Drive is here: (https://sharedresearch.jp/system/report_updates/pdfs/000/019/650/original/6324_EN_20180615.pdf?1529044314)
The R&D expense forecast includes an increase in spending related to research being conducted jointly with the Stanford Research Institute (SRI) in the US (with HDSI also in charge of mass production). This research concerns the abacus reducer (cycloidal drive).
HDSI has not included any earnings contribution from the abacus reducer in the medium-term plan. Costs and production both seem to be somewhat delayed, but the company says it hopes to reach the stage of mass production by FY03/21, the final year of the plan.
HTH
Looks just like a company wishing to avoid any competition with their harmonic drive for as long as is possible…
If someone was interested in 3D printing or CNC’ing such a device for research purposes, where would one find the best info without reinventing the wheel?