[i386]

I’m “unlocking” that this will be dead within 18 months. It will get too hard and bored engineers will head back to Google Corp where the promotion game is better.

I’m in manufacturing. Machinery is highly specialised. Making a generic robot without taking up huge amount of floor space and/or huge leaps in programming is like… Kubernetes being good for hosting your moms book club blog.

[TaylorAlexander]

I worked at Google X for two years on a different robotics effort.

In general you’re not wrong. It’s also an easy prediction to make because almost all new robotics efforts fail.

I’ll just say that they’ve been working with manufacturers for years on test projects. So this isn’t really a new thing so much as a formalization of whatever they came up with.

I’m still not sure how this stuff will go (X is of course for experimental stuff) but if the project has “graduated” to a company than I’d presume they have some revenue plan and potential customers. I could imagine that at least one of their trial projects with a manufacturer has worked out, and they think they can find more partners for this work. (I only scanned the beginning of the article)

They have been experimenting and developing a LOT of stuff since 2013 and plenty of it sadly does just get completely sacked and mothballed (and I have my personal gripes with the great projects they could have open sourced but instead they just let the bits rot). That said I think with how much I saw them working on they probably could find applications where profit is a real possibility.

[blueblisters]

What do you think of Covariant.ai? They seem to be doing fairly well in automating a lot of common assembly line tasks (picking, placing and sorting).

[foobiekr]

Pick and place was one of the very first of the modern roboticization era in the 80s. That’s more going after an existing market than this.

[danieldisu]

I'll remind myself in 2 years to see the state of this

[kybishop]

Given that this is Google, my money is on them either abandoning it, or making a new version every 2-3 years (for no discernable reason) making no actual progress as a result.

I would LOVE to be proven wrong, but Google's track record with bringing products to market, and actually keeping them for longer than a media boost, is downright depressing.

At some point it feels like every Google side project is just a media buy for their real business: hiring engineers to drive more ad revenue, engineers who are enticed by shiny projects like these.

[foobiekr]

It’s not depressing, it’s wonderful. Google’s abject failure in almost every area is great. They have nearly unlimited resources and a fantastic (though but as great as they think) engineering pool. It would suck if they managed to successfully attack adjacencies.

[kybishop]

I have a hard time calling immense wastes of resources "wonderful". Think of how many man-hours of at least better-than-average engineers have been wasted developing something like 5 separate chat apps... it's absurd.

[ta988]

Totally agree with that. There are ways to do it, but the approach they show is doomed to fail. Arrival had excellent ideas and kind of working products but if nobody injects money in them they are done pretty soon.