SoylentNews is people
SoylentNews
A good watch can take a beating and keep on ticking. With the right parts, can a rover do the same on a planet like Venus?
A concept inspired by clockwork computers and World War I tanks could one day help us find out.
[...] AREE was first proposed in 2015 by Jonathan Sauder, a mechatronics engineer at JPL. He was inspired by mechanical computers, which use levers and gears to make calculations rather than electronics.
By avoiding electronics, a rover might be able to better explore Venus. The planet's hellish atmosphere creates pressures that would crush most submarines. Its average surface temperature is 864 degrees Fahrenheit (462 degrees Celsius), high enough to melt lead.
[...] Another problem will be communications. Without electronics, how would you transmit science data? Current plans are inspired by another age-old technology: Morse code.
An orbiting spacecraft could ping the rover using radar. The rover would have a radar target, which if shaped correctly, would act like "stealth technology in reverse," Sauder said. Stealth planes have special shapes that disperse radar signals; Sauder is exploring how to shape these targets to brightly reflect signals instead. Adding a rotating shutter in front of the radar target would allow the rover to turn the bright, reflected spot on and off, communicating much like signal lamps on Navy ships.
Mechanical computers and Morse code. The future of Venusian exploration is steampunk.
(Score: 4, Insightful) by jmorris on Tuesday August 29 2017, @08:52PM (12 children)
Without a receiver this thing would need to be totally autonomous. Ok, storing a fair amount of code as bumps on metal seems plausible enough but the computer is what bothers me. Mechanical implies fairly large metal bits moving around which implies energy being used. A wind turbine is going to provide that and motive power? Ok, at those pressures any movement of the atmosphere will impart a lot of energy but that works both ways when you start trying to drive through that soup. Now consider how many moving parts would need to be stuffed in to get to minimal microprocessor levels of function, even a 4004 is a fair amount of moving parts. Clock speed is going to be very low, but that is probably ok since transmission rate is going to be a bit or two per second so there isn't much point in accumulating much data.
Good luck with navigation. How does it decide where to move? Mechanical optical sensors, radar, sonar?
Does make one wonder, could a working computer be built mechanically at a practical size / cost? Ultimate in EMP / radiation hardening, survivable, etc. Wind the mainspring and go. Somehow I doubt anything larger than MEMs would clock into KHz though. Wouldn't need to be super capable to be useful if a suitable medium density storage media was available. Think doomsday library, minimal calculation, industrial control, etc.
(Score: 2) by bob_super on Tuesday August 29 2017, @09:43PM
You're also forgetting the fun of designing the copper-clad space blimp which will take it from Earth, and deposit that thing on the Venusian surface without exceeding 0.1G at any time during the trip.
(Score: 4, Interesting) by qzm on Tuesday August 29 2017, @10:03PM (1 child)
Having trouble?
I would suggest that this is a good example of someone with a cute hobby desperately looking for an application.
There is NO POSSIBLE WAY you can build enough mechanical computation for ANYTHING useful in space exploration - it is not even distantly possible.
Sure, you could build a simple line-follower, perhaps even with enough smarts to backup-and-retry.
But a research probe? Someone has been reading far FAR too much steampunk. You could not even manage the calculations required to point a communications dish back at earth (not to mention what to then use it for..).
Its all rather silly. It is much MUCH easier to build electronics to survive than it is to build mechanics to survive, for a start.
NASA is still caught up in its modern 'we know best about everything, practicality be damned' bloat phase which has bogged down US space science for decades, until it is now being dragged slowly forward by private industry.
(Score: 1) by khallow on Tuesday August 29 2017, @11:27PM
How are you going to communicate with that dish? Sound waves? Analogue computers could handle that problem BTW. It's not that hard.
(Score: 0) by Anonymous Coward on Tuesday August 29 2017, @10:12PM
You don't need much navigation to be useful -- walking in an ever-expanding spiral is pretty good if you don't land near any obstacles, and a roomba-like algorithm (on obstacle detection: back up, rotate, and resume) is enough to get interesting coverage of an area with occasional obstacles. Not good if it gets bouncing back and forth between the sofa and the wall, but let's launch a dozen and hope only a couple are that unlucky.
The hard part is getting actual data back. Sure, you can read a thermometer and send the reading in morse code, but you'll get little benefit from mobility. The cameras, spectrographs, etc. that actually benefit from being on a rover, taking readings in a bunch of places, and looking in a bunch of directions, are much harder, and crucially take much longer, to mechanically digitize and transmit.
I think a better plan might be to use the mechanical concept for all the parts that handle a lot of power and/or are necessarily exposed to the environment, while retaining electronics for processing and transmitting data; this electronics package will require minimal power, and live in a tiny box with thick insulation deep inside the rover. It may be mechanically refrigerated, or just cooled with a tank of cryogenic gas.
(Score: 2) by Virindi on Tuesday August 29 2017, @10:17PM (3 children)
There is a simple solution: have it do no computation. All computation is done by a real computer aboard an orbiter, which sends commands of the type of "move left wheel forward" or whatever.
It would just need some way to convert incoming commands to a physical reaction, and a way to convert a physical input into a return signal.
(Score: 2) by driven on Tuesday August 29 2017, @10:54PM (2 children)
Very interesting idea. However, how would you analyze soil samples?
(Score: 2) by Virindi on Tuesday August 29 2017, @11:16PM
For any kind of sensing you'd need the ability to translate an interesting sensed quantity into a physical movement.
Imagine you had a chemical which was solid but decomposed under the presence of certain compounds in the sample. It could hold back a spring attached to a reflector tuned to a particular frequency. Then, the reaction of the reflector would indicate the presence of those compounds. Some clever chemistry would be required :)
(Score: 2) by kazzie on Wednesday August 30 2017, @10:27AM
In the same way Qui-Gon took a blood sample from Anakin and asked Obi-Wan to analyse it over the communicator. Don't worry about piddling details.
(Score: 4, Informative) by crafoo on Wednesday August 30 2017, @04:19AM (1 child)
Not being rude, but it sounds like you grew up well after the microprocessor. That's the paradigm you think in with respect to computing. Have you ever seen mechanical fire control computers on Navy ships? They calculated trajectories, in-flight times, and compensated for wind, curve of the earth, all kinds of shit.
Consider that the 3D surface profile (shape) of a freeform rotating cylinder encodes a f(x,y) parametric equation with a 3D coordinate output. Put a cam follower on it and physically dial-in the answer.
https://www.youtube.com/watch?v=s1i-dnAH9Y4 [youtube.com]
You are thinking in terms of building a generic, general purpose microprocessor with sensor inputs, A/D, and communications. But that's not what they are proposing.
(Score: 2) by jmorris on Wednesday August 30 2017, @05:15AM
Yup, I know about that stuff. I know about the old analog electronic computers too. But they are proposing sending this thing with zero reception capability, that means it has to be smarter than any of that stuff can do in the size and weight limitations of a typical rover. Optimized hardware can replace big stacks of software, usually with better performance in response time, even in the digital realm, as anyone who has dealt with embedded knows. In the end though, you need software if a machine is going to make many decisions. There is a reason almost everything has been replaced with software. Launching a few dozen and hoping one gets into a spot where a canned route won't end in a disaster isn't cost effective.
Wonder if a computer couldn't be kept alive long enough to sense the area near the lander, do the hard work of plotting out a route, drill points for samples, etc. and write all that to a mechanically readable "program" for a purely mechanical system to playback weeks / months after the electronics are fried. Then it slowly creeps along as it also slowly sends back telemetry. Just that much would be a challenge in a purely mechanical drive. Just saying turn 45 degrees left, drive 2.4 meters wouldn't be accurate enough, it would need to know how to compensate for driving over uneven ground, track slippage, etc. to avoid becoming hopelessly lost within a few days.
(Score: 2) by JoeMerchant on Wednesday August 30 2017, @01:09PM (1 child)
Bumps on metal in a highly corrosive environment, what could possibly go wrong?
🌻🌻 [google.com]
(Score: 0) by Anonymous Coward on Wednesday August 30 2017, @01:41PM
Someone in search of scrap metal might come along and take it.