Well, that about wraps it up for humanity (video: nano quadrotors)

[MattKeil]

Get your crowbars ready, wannabe Gordon Freemans, because as soon as someone weaponizes a swarm of these things, the robopocalypse is nigh.

A Swarm of Nano Quadrotors (http://www.youtube.com/watch?v=YQIMGV5vtd4&feature=youtu.be)

[John Many Jars]

fukkk

[Miramon]

Nice, but it should be "mini" not "nano" :)

[mystery]

Nice, but it should be "mini" not "nano" :)

That was exactly my reaction upon reading the description of the video. And then I watched it, and realized my opinion on grammar or taxonomy doesn't matter any more. Nothing does.

[snowcrash22]

I'm only going to speak mini-killbot gifs in this thread.;)

[ElGuapo]

The footage of them going through the window was very creepy amd sci-fi.

[Miramon]

If it was individual devices swarming intelligently based on their own on-board rules with no external commands except high level ones like "figure 8" or "head through the window to destination (X,Y,Z)" that would be one thing.

But if they're really being controlled by some remote computer, it's less impressive. After all, planning motion in mostly empty space is not hard if you have good sensors and the devices respond with reasonable precision to your commands. The feat in that case would be designing the devices such that they have that precision capability, which is somewhat less frightening.

So, how many takes was necessary for the figure 8 segment? Was it luck that it wasn't a quadrotor demolition derby, was it just a question of putting quadrotor 7 into slot 7, or do the quadrotors automatically avoid one another while still queuing into formation?

[Houngan]

If it was individual devices swarming intelligently based on their own on-board rules with no external commands except high level ones like "figure 8" or "head through the window to destination (X,Y,Z)" that would be one thing.

But if they're really being controlled by some remote computer, it's less impressive. After all, planning motion in mostly empty space is not hard if you have good sensors and the devices respond with reasonable precision to your commands. The feat in that case would be designing the devices such that they have that precision capability, which is somewhat less frightening.

So, how many takes was necessary for the figure 8 segment? Was it luck that it wasn't a quadrotor demolition derby, was it just a question of putting quadrotor 7 into slot 7, or do the quadrotors automatically avoid one another while still queuing into formation?

. . . because of all the things this video suggests, continued miniaturization and refinement of computers is the least likely. YOU FOOL!

[MattKeil]

If it was individual devices swarming intelligently based on their own on-board rules with no external commands except high level ones like "figure 8" or "head through the window to destination (X,Y,Z)" that would be one thing.

You really think that's an impossibility? Or not their next step in the project?


But if they're really being controlled by some remote computer, it's less impressive. After all, planning motion in mostly empty space is not hard if you have good sensors and the devices respond with reasonable precision to your commands. The feat in that case would be designing the devices such that they have that precision capability, which is somewhat less frightening.

So, how many takes was necessary for the figure 8 segment? Was it luck that it wasn't a quadrotor demolition derby, was it just a question of putting quadrotor 7 into slot 7, or do the quadrotors automatically avoid one another while still queuing into formation?

It's this kind of dismissive thinking that will result in your being in the first wave of casualties when the Metal Ones come for you.

And they will.

[Dave Markell]

Now, more than ever, we all need Old Glory robot insurance (http://www.hulu.com/watch/2340/saturday-night-live-old-glory).

[krayzkrok]

Aw, they're cute. I want o̶n̶e̶ fifty!

[Miramon]

You really think that's an impossibility? Or not their next step in the project?

No, I was just wondering which it actually was, self-controlled machines acting in concert, or a bunch of remote-controlled drones being shown off.

[John Many Jars]

One or fifty, the magnetic field emitted by my steely Johnson will cause them to go TOPSY-TURVY.

[Pogo]

No, I was just wondering which it actually was, self-controlled machines acting in concert, or a bunch of remote-controlled drones being shown off.

And I think his point is that it doesn't matter right now. The tech demo may be more about the flight stabilization and movement capabilities than of the programming, but with this type of electronic miniaturization it's only inevitable that sensors, communications (bluetooth or wireless, either one right now is extremely tiny) and dynamic algorithms will play together to create what is essentially an electronic flock of birds or swarm of bees, seamlessly incorporating additional units into formations while allowing each unit to react to environmental conditions via its own stabilization abilities.

Humanity = fucked, but not until these things can be powered for extended periods of time.

[Volksy]

Someone better call Tom Selleck (http://www.youtube.com/watch?v=S9zzHnRYxtc&feature=related).

[TranquilityBase]

As is evolution's way, a predator appears for these quadrotorbots.

Introducing Robodactyl!

http://img855.imageshack.us/img855/4198/cometapajaro.gif

[Jason Townsend]

. . . because of all the things this video suggests, continued miniaturization and refinement of computers is the least likely. YOU FOOL!

I lolled.

And we're clearly doomed.

[Miramon]

http://www.dilbert.com/fast/2012-01-31/

[robsam]

Why are they not selling these as toys? My tiny little indoor helicopter is fun but difficult, one of these guys (or a starter pack of ten!) would be amazing. Screw science, think of profit!

[MattKeil]

Why are they not selling these as toys? My tiny little indoor helicopter is fun but difficult, one of these guys (or a starter pack of ten!) would be amazing. Screw science, think of profit!

This is how it starts.

[ElGuapo]

Put a tiny rotating blade on each one of these.

Where is your human God now?

[Dreamshadow]

And sales of net launchers went through the roof.

[Miramon]

And sales of net launchers went through the roof.

For these things, tennis racquets should be good enough....

[Volksy]

For these things, tennis racquets should be good enough....

"Your arms will tire long before we run out of Nanocopters, human."

- Skynet

[drake113]

Someone better call Tom Selleck (http://www.youtube.com/watch?v=S9zzHnRYxtc&feature=related).

Seriously (http://abcnews.go.com/Technology/wireStory/sandia-labs-engineers-create-guided-bullet-15495497#.TysIushnDto).

[Volksy]

Well at least they haven't started to make digital actors like in Looker (http://www.youtube.com/watch?v=YKuMguUKGyg). Oh, wait.

No worries, they don't have the hypnotic flashy gun things from that movie yet. Or..do they?

/me hides in closet

[RichVR]

OMG! A bunch of little plastic flying things! Oh crap! It's the end of the world!

[Becoming]

OMG! A bunch of little plastic flying things! Oh crap! It's the end of the world!

Picture a small explosive charge built into them and a dozen swarming to your face before exploding.

Still funny?

[CLWheeljack]

Picture a small explosive charge built into them and a dozen swarming to your face before exploding.

Still funny?

I prefer to imagine them with tranquilizer darts, neutralizing resistance before the mothership comes in for harvesting.

[AlanC]

This is how it starts.

Now available in 16 colors! Collect them all!

[Miramon]

Picture a small explosive charge built into them and a dozen swarming to your face before exploding.

Still funny?

Yes, really. Each of those things must cost thousands to produce. A thug throwing a hand-grenade is much cheaper. Have to do something like the tranq dart concept that doesn't waste the machine on every use.

[ElGuapo]

Yes, really. Each of those things must cost thousands to produce. A thug throwing a hand-grenade is much cheaper. Have to do something like the tranq dart concept that doesn't waste the machine on every use.

A thug maybe, but if a U.S. Soldier tosses that grenade, now your cost per grenade is in the 5 figures.

Put another way, if we could deploy these and kill each enemy solider for a cost of say, $3000, it would be a great cost/benefit ratio. Hell, a few years ago we sent over a $6 billion aircraft carrier, plus associated crew, plus associate aircraft, fuel, etc. They killed exactly zero enemy soldiers.

Say you want to pacify a city with these things. Say their cost is $3000 each. You need the fear of say, 1000 of them to really terrorize the city.

$3 million to take over a city with these things? That's an easy decision, financially.

Give it a name like the M-107 Robot Bomb Swarm, and it's just another line item to join the already existing robots on the battlefield (i.e. the Redator and its ilk).

[MikeJ]

I don't see why they would costs thousands, once you started mass-producing them. As mentioned earlier, probably the main limit is a power source that can keep them flying for a significant length of time.

[benloran]

They just need to tweak them a bit so that the machines can mass-produce themselves, and then we can get on with the robot apocalypse.

[Miramon]

I don't see why they would costs thousands, once you started mass-producing them. As mentioned earlier, probably the main limit is a power source that can keep them flying for a significant length of time.

1.The sensors required to keep one of those things stable and precisely located at a given location must be quite a challenge. In the demo, they don't have to mounted on the device, but for any useful operation, they would have to be. I suppose the rotors themselves provide gyro stability, but for independent flight you still need some kind of fine positioning, and some kind of sensors to map and model things like windows and stairwells, and that may not even be possible to mount on such a small device right now, even putting aside computational requirements. GPS is far too coarse for determining location for precision flying, and cameras are unsatisfactory without massive remote computational support.

2. For remote telepresence piloting, you'd need a high power transmitter for video bandwidth which probably requires batteries too heavy for the device if its going to be flying for more than a few minutes. Even powering the rotors probably burns so much power that flight time is limited without a payload -- though I'd like to find out how long the demo device can run before it crashes -- maybe I'm wrong about the energy requirements.

3. The current state of AI is far from being able to guide such a device on an independent mission without telepresent operator control.

So while it's an impressive demo within a room with everything controlled by external computers and maybe even with special tracking elements mounted in the room, I don't think it will be useful for any operational purpose except possibly non-real-time reconnaissance (where it returns to the launch location to download a set of images) any time soon.

[RepoMan]

Yeah, no way those things cost thousands, even today. Hell, Amazon already has 'em for $300, though not as teeny as the little ones. Still, smaller generally means cheaper, especially once you scale up. Wouldn't surprise me if, five years from now, these could be cranked out for $100 a pop. Of course, their range is probably for shit, but still, you can do a lot of amusing damage with just (say) a quarter-mile range on these things....

[Eightball]

They just need to tweak them a bit so that the machines can mass-produce themselves, and then we can get on with the robot apocalypse.

Robocalypse!

[RichVR]

Picture a small explosive charge built into them and a dozen swarming to your face before exploding.

Still funny?

Tennis racket.

[MattKeil]

Tennis racket.

Unless you're General Grievous, one of those little fuckers is going to get through. Hell, I'd just send one with a mounted blade straight into the back of your neck while you were playing Billie Jean King with the others.

[Brad Grenz]

I want a personal swarm I can send against my enemies. Especially little kids that look at me funny...

[Nixxter]

Tennis racket.

Clearly you lose gamer cred here, sir.

Have you forgotten the lessons Gordon Freeman taught us in Half Life?

Do you not remember, like a terrible dream, this very subject of which we speak?

Allow me to refresh your memory. Behold the small yet vicious Manhack. Which came in flying packs, with sharp metal spinning blades, going for your gonads. I rest my case.

http://half-life.wikia.com/wiki/Manhack

[MatthewF]

Ghost Recon: Future Soldier has these. Before they even knew they had actually been invented. Gameplay trailer (http://www.youtube.com/watch?v=AxmH3v9_qAE&feature=player_embedded). D0med.

[RichVR]

Unless you're General Grievous, one of those little fuckers is going to get through. Hell, I'd just send one with a mounted blade straight into the back of your neck while you were playing Billie Jean King with the others.

Wow. You related to one of the guys that built them? Sorry if I twanged a nerve there.

[MattKeil]

Wow. You related to one of the guys that built them? Sorry if I twanged a nerve there.

I meant that's how I'd defeat your feeble barrier were I controlling the swarm, not that I actually want to stab you in the head. These are the kinds of situations you must take into account if you're to defeat the Metal Menace to come.

[MikeJ]

1.The sensors required to keep one of those things stable and precisely located at a given location must be quite a challenge. In the demo, they don't have to mounted on the device, but for any useful operation, they would have to be. I suppose the rotors themselves provide gyro stability, but for independent flight you still need some kind of fine positioning, and some kind of sensors to map and model things like windows and stairwells, and that may not even be possible to mount on such a small device right now, even putting aside computational requirements. GPS is far too coarse for determining location for precision flying, and cameras are unsatisfactory without massive remote computational support.

I saw a demo of a very similar system about 1.5 years ago at a conference. IIRC, all the required sensors were mounted on the device (cellphone-quality cameras?), though they had a simplified environment (grid mat on the floor) and I think they did offload the computing to a nearby laptop. People have made a lot of progress in terms of automatically mapping and modelling the environment. I think the kind of processors we will get in next-gen mobile phone or maybe the gen after that would be up to the task of handling all this stuff.

I agree that power is probably a big (the biggest?) problem.

Definitely not saying this stuff is easy to do. My hat is off to them. Once you have it set up, though, I think you could run them them off for a couple hundred or so.



3. The current state of AI is far from being able to guide such a device on an independent mission without telepresent operator control.


My sense if that 'fly through that window and blow up' would be possible, though probably not reliable enough right now. So as a kind of slightly smarter mobile sensor/munition. Definitely not hunter-killer drones or whatever.


I don't think it will be useful for any operational purpose except possibly non-real-time reconnaissance (where it returns to the launch location to download a set of images) any time soon.

I saw another demo where people were able to do pretty clever things in terms of managing bandwidth from independent aerial drones. I guess it depends what 'soon' means.

[CLWheeljack]

I assume we're talking about usefulness from an action movie super-spy perspective rather than a practical real-world perspective.

That being said, power / range isn't a problem if you just use them for short range intervention. I could foresee a briefcase sized charging / deployment device that they return to for charging, receiving orders, etc. If the target has gotten more than 1 hour or so range away, it's a job for the larger droids.

That's assuming that you're using them as targeted hunter-killers or infantry assistance. If you're talking about sentries, maybe you could have some kind of distributed energy deployment throughout your base / compound, if done properly they might not even need to stop to dock, they could be charging pretty much all the time even while they're flying.

[Houngan]

Whatever happened to Tesla's power broadcasting? That would serve nicely, you just shoot a carrier missile into a buiilding with a battery pack in the base, and it broadcasts power to the robots as they destroy all life.

H.

[Miramon]

Whatever happened to Tesla's power broadcasting? That would serve nicely, you just shoot a carrier missile into a buiilding with a battery pack in the base, and it broadcasts power to the robots as they destroy all life.

H.

It doesn't actually work, is what happened to it. Tesla was wrong about it being cost-effective; broadcast is of course a pessimal way to transmit power, and that "pumping energy into the earth's magnetic field" thing was just wishful thinking with no science or engineering to back it up, like a number of his later projects.

Nowadays, for a non-moving target, you could in theory use a maser to transmit power if nothing gets in the way; the inefficiency is still probably pretty bad, but not as bad as broadcasting. http://www.eng.tau.ac.il/research/FEL/0b0.html says someone tried this to power a small helicopter in 1964. Evidently it's too inefficient (or too dangerous if someone gets in the way of the beam) or it would have been developed further by now. That precis itself is from 1996, and so far as I know microwaved power has few or no real-world applications at present unless you count RFIDs.

[Houngan]

It doesn't actually work, is what happened to it. Tesla was wrong about it being cost-effective; broadcast is of course a pessimal way to transmit power, and that "pumping energy into the earth's magnetic field" thing was just wishful thinking with no science or engineering to back it up, like a number of his later projects.

Nowadays, for a non-moving target, you could in theory use a maser to transmit power if nothing gets in the way; the inefficiency is still probably pretty bad, but not as bad as broadcasting. http://www.eng.tau.ac.il/research/FEL/0b0.html says someone tried this to power a small helicopter in 1964. Evidently it's too inefficient (or too dangerous if someone gets in the way of the beam) or it would have been developed further by now. That precis itself is from 1996, and so far as I know microwaved power has few or no real-world applications at present unless you count RFIDs.

I read up on it in Wikipedia, looks like it breaks a number of laws of physics in most theoretical applications. Oh well, how about building the fliers completely out of supercapacitor material? I'm still waiting for that to show up.

[Miramon]

I read up on it in Wikipedia, looks like it breaks a number of laws of physics in most theoretical applications. Oh well, how about building the fliers completely out of supercapacitor material? I'm still waiting for that to show up.

Super- and hyper-capacitors do exist; in fact you can now buy one for use in phones or laptops (forget which) to replace your battery. However, while hypercapacitor power output (and input on charging) is much higher than batteries, energy capacity is much lower. So it seems like buying that product I mentioned is a dumb idea, because what matters for mobile devices is not how fast it charges so much as how long it can go between charges, and of course this is even more true of nano-quad-copter-killing-machines....

[Houngan]

Super- and hyper-capacitors do exist; in fact you can now buy one for use in phones or laptops (forget which) to replace your battery. However, while hypercapacitor power output (and input on charging) is much higher than batteries, energy capacity is much lower. So it seems like buying that product I mentioned is a dumb idea, because what matters for mobile devices is not how fast it charges so much as how long it can go between charges, and of course this is even more true of nano-quad-copter-killing-machines....

Looks like an order of magnitude less than a lead battery, and two orders less than a lithium. I bet you could still get a few minutes of flight time out of them, but what I'm reading doesn't seem to be the nanoscale carpet-like supercapacitors everyone was excited about, rather it's just a layered sandwich capacitor. Maybe when the internet becomes self-aware it can work it out for us.

[Teiman]

I think the bigger limiting factor is the fact that a militar group have not the upper hand in software. If you have a group like the USA military tryiing to invade Iran, the Iran hackers will do a lot of damage, because the Iran hackers will be better that the average sysop on the USA military. The militars know this, so are not really all that interested on creating too strong a dependency on fancy toys. Withouth that limit, I suppose is obvius a machine can kill faster and be more efficient, if you can produce a hivemind of minimachines like this one, and conquer a city, withouth losing a single soldier life, then is great. Is not great because the posibility of the enemy hacking your own drones to use the remote killswitch, or to send against you, is very very high.

[Miramon]

I think the bigger limiting factor is the fact that a militar group have not the upper hand in software. If you have a group like the USA military tryiing to invade Iran, the Iran hackers will do a lot of damage, because the Iran hackers will be better that the average sysop on the USA military. The militars know this, so are not really all that interested on creating too strong a dependency on fancy toys. Withouth that limit, I suppose is obvius a machine can kill faster and be more efficient, if you can produce a hivemind of minimachines like this one, and conquer a city, withouth losing a single soldier life, then is great. Is not great because the posibility of the enemy hacking your own drones to use the remote killswitch, or to send against you, is very very high.

What's sad is this could even be true, despite the fact it should be almost trivial to defend against any reasonable hacker.

It doesn't require a complicated security scheme, just simple crypto, with manual key exchange done before the mission, away from any network. The obvious thing to do is to make sure nothing related to a drone is on the Internet, so the only possible access is through the radio communication between the controller and the drone.

Imagine a simple non-networked hand-held box with two functions. First it generates a key pair and installs the public key into the drone by actually plugging the box into the drone for a few seconds. Second, it plugs into the output of a drone control rig, before going to the actual radio. Thus the control stream is encrypted with no security issues for key exchange, and also no security issues for Internet (or any other kind) of network access to the private key. The keys are regenerated for every sortie, so crypto hackers only have a few hours to break the keys, for a key size that is expected to require years to break. Moreover, if a drone is captured, no other drone is compromised. You could publish the details of the security scheme to the world, and it still wouldn't do the attacker any good.

But given the competence of our MIC companies, I wouldn't be surprised if the controller console is on the Internet with Microsoft zero day vulnerabilities, or if the encryption or key exchange is somehow screwed up to the point it's breakable.

Edit: Actually, why bother with 2 keys. Since you have the box and the drone in the same physical place before the mission, use a single symmetric mission key and the crypto will require less computation. This might even be simple enough for Raytheon or whoever the contractor is to implement without screwing it up.

[Miramon]

They have returned (http://www.youtube.com/watch?v=_sUeGC-8dyk).

[Jason Townsend]

Trying to lull us into a false sense of LOL.

[Houngan]

That. Was. The. Greatest. Thing. Ever.

[RichVR]

They would make great drones for target practice.

[Teiman]

If the drones itself are safe, the crackers can go against the controllers enviroment. Also it occurs to me that all cyberwars are global wars: any friend of the attacker can help, so its. Its not usa versus a tiny country, but usa versus usa and the world. Like.. If the attackers discover that the return of the data use the internet and is centralized in Las Vegas, can blind the drones by killing internet in Las Vegas by corrupting the routes.

[Miramon]

If the drones itself are safe, the crackers can go against the controllers enviroment. Also it occurs to me that all cyberwars are global wars: any friend of the attacker can help, so its. Its not usa versus a tiny country, but usa versus usa and the world. Like.. If the attackers discover that the return of the data use the internet and is centralized in Las Vegas, can blind the drones by killing internet in Las Vegas by corrupting the routes.

Why would any military use the Internet at all for any system whatsoever associated with these drones? That would be incredibly stupid. If they do use operators at a far distant installation (which is very dumb in itself, but supposedly they do), then they must use a beamed uplink to a military communications satellite to get around the globe.

[TranquilityBase]

Who would've thought that the rise of Skynet begins with replacing OK GO?

[Teiman]

Why would any military use the Internet at all for any system whatsoever associated with these drones? That would be incredibly stupid. If they do use operators at a far distant installation (which is very dumb in itself, but supposedly they do), then they must use a beamed uplink to a military communications satellite to get around the globe.

About the use of internet. I don't know, but at some point could be cheap, or required if the bandwidth of other methods is not enough.

About location in Las Vegas, it may sounds unlikely, but a lot of the drones currently operating in afganistan work from there:

http://www.theatlantic.com/magazine/archive/2006/09/hunting-the-taliban-in-las-vegas/5116/

And you are right, are using uplinks to satellites. But perhaps a army of small drones would need much more bandwidth or better latency.

[Miramon]

Enter the sand flea.

http://www.youtube.com/watch?v=6b4ZZQkcNEo

Boston Dynamics is pretty cool. They've been doing some amazing stuff over the last few years.

[CLWheeljack]

Enter the sand flea.

http://www.youtube.com/watch?v=6b4ZZQkcNEo

Boston Dynamics is pretty cool. They've been doing some amazing stuff over the last few years.

That reminds me of Speed Racer's hydraulic-jack assisted jumps. But you know, real.

[Houngan]

That seems like a damned good idea. I'm guessing some sort of worm-drive loading a spring striker that's fired out the back. It's always great to see a simple solution to a complex problem.

[Miramon]

That seems like a damned good idea. I'm guessing some sort of worm-drive loading a spring striker that's fired out the back. It's always great to see a simple solution to a complex problem.

Yeah. And the resilience is nice too; the wheels don't fly off and the chassis doesn't shatter or anything :) Of course it's manually remote controlled, not autonomous, but even so it's a very nice little demo.

[Houngan]

Yeah. And the resilience is nice too; the wheels don't fly off and the chassis doesn't shatter or anything :) Of course it's manually remote controlled, not autonomous, but even so it's a very nice little demo.

True, but it's not like they can't fit the quadrotor brain into that thing without it even noticing. Put my vacuum's laser-mapper on it and away you go!

Question for the smarties:

How much energy can you get from a microwave (or other EM freq.) transmission? Wouldn't it seem like a good idea to park a satellite over Mars with a solar panel and use that to charge rovers, or is there just not enough juice? I'm assuming that you would use it to charge for two days in order to get an hour of activity or some such, not that you could just run a rover from a constant stream of magic energy.

H.

[Miramon]

True, but it's not like they can't fit the quadrotor brain into that thing without it even noticing. Put my vacuum's laser-mapper on it and away you go!

Question for the smarties:

How much energy can you get from a microwave (or other EM freq.) transmission? Wouldn't it seem like a good idea to park a satellite over Mars with a solar panel and use that to charge rovers, or is there just not enough juice? I'm assuming that you would use it to charge for two days in order to get an hour of activity or some such, not that you could just run a rover from a constant stream of magic energy.

H.

I think it's possible; however you would need a relatively large solar array to power the beam, not to mention pinpoint accuracy if you're going to hit a tiny mobile antenna with a narrow maser beam -- the usual idea of solar satellite array is to use a very large farm of ground antennas covering many hectares.

I'm guessing the kind of payload that the current orbiter has is too low-mass to accommodate a large enough solar array to power a useful beam, but you could imagine either a bunch of such payloads linking up, or just a bigger delivery to Mars orbit.

Possibly in Mars' thin atmosphere, you can do better with a narrow beam than on Earth, where if you pack too much energy into a narrow beam, it ionizes the air into a plasma tunnel and then the beam path can get blown by the wind any which way. Some early US Army H-F chemical laser experiments were continuous mode, not pulsed, so when they burnt tunnels in the air they could get embarrassing results like having the beam path curved into weird loops by the wind. I guess those planetary dust storms would cut off power, but I don't think they're that common.

[Houngan]

I think it's possible; however you would need a relatively large solar array to power the beam, not to mention pinpoint accuracy if you're going to hit a tiny mobile antenna with a narrow maser beam -- the usual idea of solar satellite array is to use a very large farm of ground antennas covering many hectares.

I'm guessing the kind of payload that the current orbiter has is too low-mass to accommodate a large enough solar array to power a useful beam, but you could imagine either a bunch of such payloads linking up, or just a bigger delivery to Mars orbit.

Possibly in Mars' thin atmosphere, you can do better with a narrow beam than on Earth, where if you pack too much energy into a narrow beam, it ionizes the air into a plasma tunnel and then the beam path can get blown by the wind any which way. Some early US Army H-F chemical laser experiments were continuous mode, not pulsed, so when they burnt tunnels in the air they could get embarrassing results like having the beam path curved into weird loops by the wind. I guess those planetary dust storms would cut off power, but I don't think they're that common.

Thanks, sounds like it's a viable idea. The rovers have been so impressive I hate to see them grounded by their failing solar arrays, seems like setting one up outside of the atmosphere would be a good solution. I would also think NASA could come up with a rover delivery system that is little more than a big ol' missile the shuttle replacement could fire at Mars since they've got the rover itself licked. Just send a stream of them with some sort of uplink to find and target the maser and we could have the planet crawling with them in a few years. If you don't have all your eggs in one basket it wouldn't be prohibitive to just launch a dozen of them in hopes that three or four would make it to the surface.