Oz Blog News Commentary

Why did Google dump Boston Dynamics?

March 19, 2016 - 08:50 -- Admin

Boston Dynamics pioneered walking robots, on both two legs and four (see video). BD's early demonstrated success led Google (aka Alphabet) to acquire them for their robotics mashup, Replicant, in December of 2013.

However, Google has changed their minds.  Google is now putting Boston Dynamics up for sale.  

The business press says that the reason for this sale "was a reluctance by Boston Dynamics executives to work with Google’s other robot engineers in California and Tokyo and the unit’s failure to come up with products that could be released in the near term."  In other words, a culture clash.


My friend Monica Anderson, a pioneer in deep learning (and a Google alum), has a much better explanation.  The team at Boston Dynamics wasn't able to make the transition to deep learning, the approach Google uses to produce limited AGI (artificial general intelligence).

Boston Dynamics (BD) are very successful pioneers. But their algorithms are not based on Deep Learning (DL) principles. And Google is leading the world in Deep Learning and can apply it to anything they want, including robotics. DL based algorithms do not provide a complete robotics solution today but there is wide agreement that this is the best path forward for the field. Why is this important? The difference is robots that can walk vs robots that can dance ballet. The goal is "graceful locomotion" which will be an order of magnitude more adaptive, more energy efficient, and faster than the current generation of robots.

I am suspecting this is part of the culture clash. Imagine a meeting where Google comes to BD staff and says "we want you to toss out all the software you have written, retrain your engineering staff to use DL and other Holistic methods, and create a new generation of robots capable of learning rather than being programmed"... then I can see BD staff saying "Ahem. That'll take a few years". And Google might then say "Forget that".

Changing one person's mindset and stance from Reductionist to Holistic methods is a multi-month to multi-year effort; I have tried that and have rarely succeeded. Now imagine a company's worth of engineers that need to do that switch. It might be much better, cheaper, and faster to start a company from scratch and hire a more receptive crowd of engineers. And seed them with Googlers that already drank the Holistic Kool-Aid.

The hardware solutions BD created is know-how and intellectual property that Google currently owns since they own BD. When they sell off BD they could exclude the key patents or sell them as non-exclusive licenses. Which means Google could start a competing robotics company, use the BD patents and hardware know-how, and add a Deep Learning based software stack to run that hardware much faster than they could turn Boston Dynamics around.