Microbots Are on Their Way | Hacker News Reader

Microbots Are on Their Way

nytimes.com

101 pointssmb11137 comments

20 years ago, when I was a bright eyed graduate student, I was mesmerized by MEMS (micro-electro mechanical systems), which promised similar revolution. I learnt about 'artificial muscles', and MIT Technology Review even ran a cover issue on how MEMS will revolutionize everything. This article could have been written in the year 2000 except it would have mentioned MEMS then. Now there is no mention of it. I'm older and saner now. Still feels very much a academic pipe dream than real engineering. I dreamed of working with Kris Pister and now he is 20 years older. Another young professor at UPenn and Cornell is trying to get tenure....call me cynical but this too shall pass. Issues of toxicity in human body etc are huge...

Being wrong is one thing, being too early is an entirely different matter. The timescales of these types of innovations are very long. Early failures don't always mean it's a bad idea.

The only foolish part of your story is over-investing (and over-selling to the media/public) too early before you have any tangible results.

MEMS have been much quieter in their influence as it turns out. Now they're in everything with an IMU or accelerometer, especially including things like your phone; they're in disposable pressure sensors; and they're in microphones[1].

They've revolutionized some parts of how we live our daily lives, though not in the same way innovations like the car, airplane, computer, or cellphone have.

I expect microbots will be similar. After a decade or two of hard work and billions of dollars invested, they will quietly revolutionize some other small parts of our lives. Meanwhile, the rest of the world moves on.

https://en.wikipedia.org/wiki/Microelectromechanical_systems...

MEMS is also used for laser beam steering (depth sensors, projectors), oscillators and even loudspeakers. MEMS is truly a breakthrough where physics meet electronics.

Commercially MEMS is also very interesting because it’s a branch of semiconductor manufacturing which is dominated by different players compared to the regular TSMC/Samsung/Intel trifecta.

Some devices use MEMS oscillators instead of crystals, which has the bizarre side effect of making them allergic to helium: https://ifixit.org/blog/11986/iphones-are-allergic-to-helium...

For me enabling cheap accelerometers so it's fairly trivial to give our machines sense of 'up' and movement is a sufficient enabled advancement for any hyped up technology.

I feel like I've been seeing stories like this, with accompanying microscopic footage, for decades. Has anyone ever built anything more complex than a simple actuator? These things seem to barely qualify as robots.

I think the main thing I always see with posts like these are that they always say what could be done but with no plan or process to do any of it. The problem with microbots seems to be a similar problem to spaceflight currently: we can make it and its possible but no current commercial opportunities exist to push the technology past the "we can do it" stage.

Well, spaceflight is a big industry though. Or do you mean human spaceflight?

Anything outside of telecommunications or earth observation. Human spaceflight is a different can of worms with the additional dilemma of that no large hardware failures can occur. Rockets fail a lot, people dying is bad for PR, and thus you need a really good reason to send a person. The thing then is what do you do with people in space that makes enough money to warrant it? Basically nothing. Thus the lack of any human spaceflight other than research purposes.

I think theoretically, you could make money refining rare minerals from asteroids (like platinum, gold, rare earths). The issue is that since the startup cost is so high, the mission is very risky (conservatively, it would probably cost in the 10s of billions to be able to refine anything at scale) and 10s of billions worth of rare minerals won't stay worth 10s of billions since the world only needs so much minerals - the price would fall substantially.

I think the only thing really worth doing in space, economically speaking, will be energy related. Maybe harvesting helium 3 will be lucrative. Maybe (I doubt it) there will be a profitable way to harness solar energy - could be more profitable if we can produce the panels in space by the asteroid they are procured from, but then the issue is transmission.

Actually, there is another thing. I think space tourism could be lucrative. Imagine if you set up a lunar colony for 20b that could house ~1k people with about 200 permanent staff. If 800 people are paying $100k/week (plus cost of transportation) to stay up there, you're making $4b per year. Personally if I were worth in the ~10s of millions I would absolutely shell out $100k to spend a week on the moon so I think this kind of thing could work

It is still only a playground for the rich and wealthy. The Space industry needs to become like the Airline industry in minimizing the accidents and making it safe and cheap enough for the average person to afford.

There are people who want to go into space to be like Captain Kirk or some other scifi character. Some of them are rich or wealthy and can afford the $100K a week on the Moon.

Thing is a moon-base has to avoid meteorite strikes and other hazards. If the water or oxygen gets contaminated that's it for everyone.

>helium 3

Breakeven D-T fusion has never been demonstrated, at any scale. ITER won't even attempt D-T fusion until 2035, and then you need to figure out how to scale the reactor down until it doesn't cost $50 billion and 50 years to build, and also increase the collision energy until it can actually do D-He3 fusion, or even higher, true aneutronic He3-He3 fusion. Yes, yes, Stellarator, Polywell. Neither have been built.

Or you can just do solar and batteries. Which exist today.

>space based solar

Tom Arnold did the best post on SBS back in 2012: https://dothemath.ucsd.edu/2012/03/space-based-solar-power/ Depending on what downlink power density you're willing to tolerate, how big of a transmit dish you can assemble in orbit, and how big of a receive array you want to build on the ground, you only get maybe four times the power density of the sunlight falling on the ground taken up by the rectenna array.

Awful lot of work to go to just to get four times more power per square meter on the ground.

And if you've got the room for a rectenna array 4 km across for SBS, you might want to just build a regular terrestrial solar array 8km across instead, and save a few dozen billion dollars in launch costs.

I would think that the most lucrative thing would be to create permanent space colonies for

1. Sending most of the people there instead of destroying slowly but surely our environment here

2. As an insurance against point 1 failing.

That's true. I remember seeing that kind of thing on it's back, basically feelers in the air that could manipulate items, a really long time ago.

It couldn't have been that many years after Drexler's book got published.

I really like the slaughterbots: https://www.youtube.com/watch?v=HipTO_7mUOw

What a thought-provoking short video, thanks for sharing.

As for the researchers aiming to ban autonomous weapons at the UN, how would instituting a ban actually solve anything? It wouldn't prevent someone in their basement from building them, nor would it be a deterrent to states with a strong military from developing them.

Interestingly, black powder era technology easily defeats "stochastic motion." It's called a shotgun.

Can you defend yourself from a swarm of 10 coming from all directions at 30mph?

With 10 shotguns using the same AI? Why not?

Along similar lines:

https://www.youtube.com/watch?v=1n9BoDpnhmk

This is fantastic, but another approach would be to use the things that are already small and learn how they work, and then maybe reprogram them (i.e. bacteria).

Also, as Richard Feynman pointed out (https://www.youtube.com/watch?v=4eRCygdW--c) at small scales water is thick like honey so it's probably more efficient to use a rotating turbine mechanism (i.e. flagella) for propulsion rather than trying to shrink paddles down to micron sizes.

And almost foreign material that you put in your body will eventually be covered in bacterial biofilms. So might as well learn how to program and control the bacteria in the first place.

From the article:

> Challenges remain. For robots injected into the brain, lasers would not work as the power source. (Dr. Miskin said magnetic fields might be an alternative.) He wants to make other robots swim rather than crawl. (For tiny machines, swimming can be arduous as water becomes viscous, like honey.).

I feel like I've been reading this article every five years since 1990.

I wonder what kind of "brain" would even be possible to put in a robot this small, given the physical limit we're approaching. Could limit the potential for certain dystopian scenarios.

The ones in the article look like they're basically photocells which get activated by lasers to move the 'legs'.

At that scale, you probably couldn't have much battery power either. Maybe it'd be possible to power a small microcontroller off of radio signals which also send instructions. They wouldn't need to be very "smart" as long as something else in the room was, like a phone or router or something.

When I looked at the animated picture I wondered about the flashing dots on the circuit till I realized that these are probably reflections of the laser beams directed onto the microbot.

A cluster brain/Hive mind would be one option and splitting the brain across several robots.

However - No need to have the computing/brain in the robot at all. Equally a group of these robots sharing sensor data would make some central processing design even more suitable and allows the robots to not be limited by just their own sensors. So many upsides with that aspect.

Onboard you would need very limited processing and wireless comms and that would be it, all the intelligence can then be offloaded and would also make such robots cheaper and less energy dependant. More so as with lower power usage, wireless powered becomes much easier, as well as offer a wireless communication channel using ambient backscatter.

A simple one. With robots this size, you'd want to rely on numbers, not smarts. Kind of like life does at cellular and sub-cellular level - things don't think much, but they sure do bump into each other frequently, and there's a lot of them.

This looks like the most important point:

> Dr. Miskin worked around the power conundrum by leaving out the batteries. Instead, he powers the robots by shining lasers on tiny solar panels on their backs

I've played the Metal Gear Solid games enough to know the crazy possibilities of nano machines.

"could one day", beware these three word in a title.

I can't tell if it's cute or absolutely terrifying.

Meanwhile, https://youtu.be/vOLvFhdkLmA

Flying microdrones have obvious applications (entertainment, replacing fireworks, on demand traffic guidance, etc.), but ground based versions are a mystery to me...

Does anyone else look at the microbot next to the paramecium and see an Atari sprite next to a photograph?

Or maybe we could inject them into the brain, and then could 'override basic autonomic function. Maybe we could use this on a recently dead body and it could do simple things, like amble around, maybe grab things, or bite.

Reply on news.ycombinator.com