BioRobots Who Build Themselves
How colonizing other worlds and rebuilding our own require an entirely different set of technologies.
Tesla called it the “machine that builds the machine”. Viv created software that began to program itself. Bloomberg even did a video story on a spider-like robot that 3D prints itself. What does this all add up to?
We’re beginning to see an emergence of software and hardware who’s platform its built upon is, well, itself. The kind of rapid, exponential advancement that does not require human intervention to continue refining and prototyping is what’s required for what we would include as real “general” intelligence.
Much of the AI industry likes to talk about only the brain. Only the software. But if we’re truly going for biomimicry, then shouldn’t we also be concerned about the body? The hardware?
The simple term for all of this is Biorobotics. It’s the study and building of physical systems that attempt to recreate Biologic Intelligence.
II. Three Innovations of the Human “Robot”
There are three aspects of the human species that enabled it to rise above the rest of the animal kingdom into the Apex explorer (we deem the “predator” term an obsolete mechanism for a more barbaric society). There are, in order:
- Intelligence: the ability to reason, rationalize, and solve problems in isolation and as a group.
- Opposable Thumbs: the ability manipulate 3-dimensional objects in the 3-dimensional world that we occupy.
- Mobility: not just physical mobility of moving ourselves from place-to-place and objects from place-to-place, but also the ability to control our mouth musculature. This is important for the last aspect of #1 above. Communication. What started as oral communication (i.e., talking) over time moved to written communication (i.e., glyphs using our opposable thumbs).
A critical point to understand about any AI startup that you may come across is that true general intelligence will not emerge from a strictly software space. It needs to exist in both the real world and the world of thought for it to function properly. Unless, of course, you believe that matter doesn’t really exist and we are just perceiving a projection of that matter in an isolated, solely conscious space. But I digress.
We are 3-dimensional, conscious beings. We see, smell, or touch something (sensory input), decide how to react to it (intelligence), and then do something about it (motor output). Whether it’s talking or writing about our ideas to pass knowledge and understanding along to others (education), playing games with physical objects (sports), or making the world around us better (society), you can’t achieve any of these things without those very important three points.
Thinking. Building. Moving.
So why are so many other startups getting funded that only address one of those three things? Lack of vision, a plan to get there, or a payoff within the constraints of an LP-backed fund? Sure you could probably point to a few of those things.
But I think the real issue is that very few people in this world other than, say, a few well placed research scientists even know that this is possible. When you spend your time on the fringes, you’re very quickly able to internalize something new and whether it matters or not. After years thinking about next-level AI and reading about Connectomics, the minute I heard it was being deployed in real-world robotics, I knew we had crossed a very real chasm from research to commercialization.
When you’re pushing the limits of what’s possible, it requires human ingenuity on the bleeding edge of science to will it into existence.
III. What’s Intelligence Got To Do With It?
The other part that the AI startups and researchers of the day haven’t yet unlocked, but biology already has, is the temporal aspect as well as the spatial. It doesn’t just matter that the electrical circuit, or gate, gets opened, or in what order it gets opened, but it also matters how strong the electrical signal through that gate is and the amount of time it takes for a signal to propagate through a series of these gates.
As you can see, we are talking about something more fractal in nature than a simple linear regression line that works back-to-front, even if they are pulled together in series in a deep way.
Today’s cutting edge deep learning algorithms are being used by major tech companies like Google to train robotic grippers to pick objects up from a bin. You might have seen the video.
But the problem with systems like this is it represents a very small subset of what the mechanical “opposable thumbs” of humanity can do. It’s a vertical system, reach down, attempting varying combinations of movements until something works. Just pure brute force attack.
That’s really all deep learning is. It’s a brute force attack at optimizing some problem. For some nails it represents the perfect hammer. But in others, where chaos is the norm, not the exception, it fails to work properly. Like a brute force attack for a 1028-bit encrypted password, it would take longer than the life of the Universe to solve the problem.
But if there are some problems that can’t practically be solved by a brute-force attack, is there another way?
Well, we are biological organisms creating biological problems and so shouldn’t we at least try to attempt solving them using biological methods?
III. Introducing BioRobotics
The technical term for this is BioRobotics. At its most basic, it’s about mimicking biological systems using robotics. Not even Wikipedia has much depth in describing it because it’s so new. They way they define it, however, is more about mimicking the external aspect of a biological animal, but not so much the intelligence aspect.
As we mentioned above, it requires both the brain and the body to be truly biomimetic. You need Biologic Intelligence and BioRobotics working in tandem along with some user interface for human interaction.
So lets get this out of theory and put some meat on the bones. How do we bring together these 3 pieces? Funny you should ask.
- The Lieber group at Harvard are already working on the neural lace, as we’ve written about previously.
- Gary Fedder at CMU is already working on microbots, as we’ve written about previously.
- He’s working with a number of researchers at Case Western’s “biologically inspired robots”.
Are you getting it?
- Subscribe to the Newsletter
- Follow the Robotics channel
- Follow Humanizing Tech on Apple News
- Follow Humanizing Tech on Facebook
from Stories by Sean Everett on Medium http://ift.tt/2fKAj7m