When we consider many of the Tesla manufacturing innovations, one feature common to many is:That's very interesting, I've honestly never considered the possibility that humanoid robots could work FASTER than humans. More hours per day for certain, but faster?
And truly right now I'm wondering why I never considered that before, seems reasonable upon contemplation...
Reduced downtime, e.g.:
-Berlin paintshop that allowed ad hoc color changes, eliminated most primers, with water reduction, manual work reduction. That all added to faster, thus cheaper.
-the factory OS stages options through integrated software, the only 'common' feature being applying the vin as master control. Nothing manual, all automated. Faster, fewer errors, more precise thus cheaper.
- following that logic consider Optimus. By increased task precision, near-total absence of fatigue ('near' because there will be charging and maintenance) even with no increase in hours Optimus ends out being faster. Error-free virtually always is faster than error correction, plus cheaper because fewer inspections are required precisely because of the error-free process.
Reduced parts count coupled with reduced process steps, e.g
-GigaPress- by eliminating hundreds of parts, the manual processes are mostly eliminated,
process control becomes central to success and indispensable to cost reduction,
legions of Kukai work relentlessly doing precision repetition and handling cumbersome and/or heavy
processes that humans could not do.
So really, of what use is a humanoid robot when other fully automated solutions are producing such stellar results? The common thesis is that they will do dangerous or repetitive tasks that humans now do and conventional robotics cannot do, at least not economically.
I'll suggest that their primary role will be to do fairy simple tasks and/or dangerous ones that humans do because conventional robotics can't complete and entire task, that including some categories of parts loading in assembly lines, doing irregularly scheduled control functions and the conventional assembly line actions that, in older factories have not yet been automated. They even could preform a wide variety of precise production quality checking that humans do today.
My thesis is, basically, that Optimus likely requires less training than do humans to replicate many tasks. Specifically if the task involves measurements or checking for non-compliant elements, Optimus could do such ad hoc tasks far faster and more precisely than could any human.
Most Optimus task discussions and even demonstrations should it doing routine and simple tasks not really demanding any significant level of skill, just some manual training. What if all that is a red herring, that the primary industrial mission is to help improve quality control, among other things.
That does not in any way detract from the attractiveness of 'Robbie". I'd pay well to replace my household staff with Optimus. Optimus, at least, can follow instructions. The human equivalents often do not have enough education to do that, even when they want too. As for faster...with boring and repetitive things humans are really slow and error-prone. The quality of those jobs really goes up with a humanoid robot, not least because they will not do something they are not instructed to do. That alone makes them far faster than humans.
The analogy with Apple VisionPlus might be apt. Presumed to be for entertainment one solution mentioned by Tim Cook was an innovative solution by SAP. Those who know SAP know how boring and fear-inducing those interminable SAP training sessions can be. Turn much of those into immersive video! Just search for SAP Certification training and it becomes instantly evident that video games may not be the actual lucrative market for some time to come. In that respect it might be analogous to iPad, the value of which is enormous in Airline Pilot flight bags, manus and wine lists, sales and order processing etc.
With Optimus nearly everyone thinks about how it has been publicly positioned rather than what those skills and abilities might be to extract maximum value. Optimus is a long way from delivering on all that promise, but...how efficient has training become?
With no special facts, with only logic regarding analogous situations, I think we may have been too linear in our thinking.
Just think of Gigapress, >100 sold but after Tesla nearly all of Chinese Tier One, if the rumors are correct.
The huge problem is thinking about the 'installed base' assuming nothing will change.
