A Camera Drone for space

[IainMcClatchie]

Thanks for the feedback Jim.

A: Trajectory changes after midcourse corrections seem expensive.  The drone could diverge so far from Dragon that it comes down hundreds or thousands of miles away.  That could make finding it expensive, and if it slams into the ground there's a good chance the flash storage and beacon are toast.

Popping out a second drone for reentry pics seems reasonable, and much easier than recapturing the drone before manoevering.

B: No need to aim, just avoid rotating too fast.  The cameras point in all directions.  During reentry, the cameras point in most directions, just not close to forward.  How about the drone re-enters slightly behind the capsule, with a slightly smaller ballistic coefficient so it falls back as it goes?  Give it some lateral velocity before entry, so the capsule stays off to the side a bit and still in view.

C: I don't know what keeps the drone apart from the Dragon.  This seems like a deal-killer.  If the drone is the same shape as the capsule, with a slightly smaller ballistic coefficient, and released trailing (and with some lateral velocity) would that be enough?

D: Here's a video looking out from Orion during reentry:


The blue curve of the Earth is visible through the entire sequence, just barely for about a minute.  I think visibility is not a problem.  I'd probably choose a camera with fairly large pixels and 14b A/D to get good dynamic range, so the blue Earth can make a nice looking background for the fiery capsule.

[Jim]

A: Trajectory changes after midcourse corrections seem expensive.  The drone could diverge so far from Dragon that it comes down hundreds or thousands of miles away.  That could make finding it expensive, and if it slams into the ground there's a good chance the flash storage and beacon are toast.

B: No need to aim, just avoid rotating too fast.  The cameras point in all directions.  During reentry, the cameras point in most directions, just not close to forward.  How about the drone re-enters slightly behind the capsule, with a slightly smaller ballistic coefficient so it falls back as it goes?  Give it some lateral velocity before entry, so the capsule stays off to the side a bit and still in view.


D: Here's a video looking out from Orion during reentry:

A.  Any maneuver of the Dragon is going to separate them.  Even Dragon attitude changes

b.  Aiming is required.

d.  That is looking up through the "wake" and not out a side window.  this is equivalent of looking out the upper windows of the shuttle.  (notice that there are few videos of looking out the front windows for the shuttle)  Like I said there not going to be able to image through the plasma.

[Lars-J]

A.  Any maneuver of the Dragon is going to separate them.  Even Dragon attitude changes

Isn't the bigger risk that an attitude change or mid-course correction will actually cause re-contact?

The safest way to do it would be to give it a push so that it keeps drifting away... this just means that you have to pick the best time to do it. (earth-rise?) But it seems more risky than necessary.

[JamesH65]

Make drone disposable.

Small CO2 cylinder and 6 nozzles, use fuel injector style valves to do the squirting.

Raspberry Pi would have enough oomph to do the math for orientation, could also handle the camera, wifi and you can  plug in an accelerometer and GPS module for location.

Small LiPo battery.

Cheap and cheerful, total weight about 250g.

Of course, the Raspberry Pi case for the Pi's on the ISS cost £20k to make, so space stuff tends to get expensive quickly.

Lots of SW out there for controlling drones.

Hmm, might try and knock one up!

[Jim]

Raspberry Pi would have enough oomph to do the math for orientation, could also handle the camera, wifi and you can  plug in an accelerometer and GPS module for location.

Are you talking about just entry or the whole mission.

RPI would be useless since GPS isn't going to help with orientation or location at the moon.  Neither is the accelerometer.

For entry, GPS is not going to help because of the plasma.

At any rate, accelerometer and GPS is not going to provide enough info to point the camera at the capsule.

[toruonu]

For orientation towards the dragon wouldn't a regular ping from the dragon be enough to choose which way to point, maybe a slight offset two-ping system to determine precice direction in relation to the drone, then just aim towards the beacon...

[Jim]

For orientation towards the dragon wouldn't a regular ping from the dragon be enough to choose which way to point, maybe a slight offset two-ping system to determine precice direction in relation to the drone, then just aim towards the beacon...

That would require multiple antennas.  But it wouldn't work during entry

[IainMcClatchie]

Jim,

B. Pointing

I'm not sure where you are getting the pointing requirement.  The idea is that the ball of cameras has full coverage of, at the Moon, the entire sphere, and during re-entry, a bit more than half a sphere.  The cameras will still be sensitive to rotational blur, but rotation can be sensed with a $5 MEMS rate gyro and cancelled with either an internal motor/flywheel (or cold gas jets, but those raise the spectre of collisions).

If the cameras are seeing in all directions, they are seeing the capsule regardless of where it is.  This is just like the StreetView R7 camera.  (I was the tech lead for that.)

Distance is another problem, however.  The cameras will only produce a useful image out to about 1 km.  I'm not sure it's a good idea to have the drone try to manoever.  Maybe the drone can be made small enough, and the capsule can eject one every 60 seconds during re-entry.  The redundancy would be good for finding most of the footage after the things fall into the sea.

C. Separation

If the drones are design to trail and laterally separate from the Dragon during re-entry, they will be behind the capsule when it pops its chutes.  Yikes!  (Awesome visual, of course, if collision can be avoided.)

Would it be possible to eject drones which had a larger ballistic coefficient than the Dragon during re-entry?  They could then look up and sideways at the Dragon, and they'd be ahead of it when it pops chutes.  No more rain of bricks.

D: Plasma optical density

http://mms.businesswire.com/bwapps/mediaserver/ViewMedia?mgid=173435&vid=5

Looking out the side just doesn't look like a problem.  Obviously the cameras won't be staring through the heat shield.  Note that the side-mounted cameras can look forward to some extent.

[Jim]

B. Pointing

Maybe the drone can be made small enough, and the capsule can eject one every 60 seconds during re-entry. 


C. Separation

If the drones are design to trail and laterally separate from the Dragon during re-entry, they will be behind the capsule when it pops its chutes.  Yikes!  (Awesome visual, of course, if collision can be avoided.)


D: Plasma optical density

http://mms.businesswire.com/bwapps/mediaserver/ViewMedia?mgid=173435&vid=5

Looking out the side just doesn't look like a problem.  Obviously the cameras won't be staring through the heat shield.  Note that the side-mounted cameras can look forward to some extent.

B.  from where?

c.  if the drone is behind the capsule, then side looking cameras are going to be useless.

There is no way of keeping the drone within range of the capsule.

d.  It is problem.  You looking from the wrong perspective.  The sheath is thicker and basically flowing over the lens or close to the lens. 

Video of entry is not feasible. Period.

[IainMcClatchie]

Jim,

B: The drones would be attached to the top of the Dragon and would eject from there.  I have no clue how that might affect aerodynamics during launch, maybe they'd need an aero shield around them.  The Dragon has a cover over the port used to dock to the ISS.  Maybe some of that can go, given that it's not docking to anything.  Or maybe the drones can fit behind a bulged version of that cover.

C: The collision issue seems the worst of them to me.  I don't much like any of the three options:
* If the drones are ejected from the top of the Dragon, and then have a lower ballistic coefficient, they leave the Dragon vertically.  Getting enough horizontal velocity that the Dragon can be seen horizontally might not be possible, and looking down through the brightest part of the plasma sheath might also not be possible.
* If the drones are ejected from the top of the Dragon, and then have a higher ballistic coefficient, they want to plunge past Dragon.  Now the optics problem is easier, but the aero problem is dangerous.  Somehow the drone is going to get through the recirculating plasma on the side of the Dragon.  You'd need enough lateral momentum to definitely overcome any lateral aero forces during transit of the Dragon's wake, which basically means you'd need a little rocket.
* You could have the drone fly itself, varying it's drag coefficient or lift vector to try to maintain some separation from the Dragon.  Just knowing reliably what the relative location of the Dragon is seems daunting.

Maybe the drone could eject with some small aero device that makes it have a lower ballistic coefficient than Dragon, get free of the Dragon's wake, and then drop that aero device so that it now has a higher ballistic coefficient and can get out in front.

One more problem: the Dragon has a lift coefficient of around 0.5 G during re-entry.  If the drone can't match that, it's going to be separating really fast.  To match it, the drone would have to know it's own orientation for about a minute... which is possible with a reasonable set of rate gyros.  But if the drone cannot exceed the Dragon's lift, the background to the Dragon will be the black of space.  You really want a shot that shows the horizon, which requires matching lift reasonably well.  Maybe each drone, when ejected, only gets about ten seconds worth of useful imagery, and there is not continuous high resolution imagery of the Dragon exterior on the way down.  That's actually okay.  The drone imagery can be cut together with interior shots later, and for a highlight reel you only want a minute at most of re-entry, which generally takes several minutes.

D: I'm unimpressed with your bare claim.  In the Orion video, the plasma sheath pretty clearly has a radiance peak about a minute long.  It's possible that there will be a one-minute-long photography blackout, although I'll note that during the ENTIRE SEQUENCE, it's possible to see the limb of the Earth, right through the upper portion of the plasma sheath.  So at some point, off peak, it's possible to see laterally.  It's just a matter of how much of the re-entry can be shot.

Shooting through glowing fluid is just a problem of opacity and signal to noise.  Satellites image the ground through the significantly brighter blue sky all the time.  People stare at logs in a fire, right through the flames.

An immersive VR video of reentry, in which the enveloping plasma blocks out your view of the nearby Dragon at its peak irradiance, would still be an amazing experience and have great PR value.  Heck, SpaceX could try out a few drones behind a re-entering second stage, so we could see what it looks like when something breaks up, up close.  Trying it out on an unmanned Dragon coming back from ISS would be a way to validate that the drones don't hit the Dragon, and see how long they pace it close enough to get a decent shot.  I'm not sure if this last bit would work, as it might interfere with NASA's primary mission of returning samples.

[Jim]

Jim,

B: The drones would be attached to the top of the Dragon and would eject from there.  I have no clue how that might affect aerodynamics during launch, maybe they'd need an aero shield around them.  The Dragon has a cover over the port used to dock to the ISS.  Maybe some of that can go, given that it's not docking to anything.  Or maybe the drones can fit behind a bulged version of that cover.

They are not going to put it on the front.  The whole thread started with putting them in the trunk.  You just don't go and add something willy nilly.  There aren't services like power or data all around the exterior of the Dragon to separate an attached object.  Also there is the safety implications of non separations with the pressure tanks of the drone.

[IainMcClatchie]

How does dragon open that front hatch cover without power?  Is it just a cable like the trunk release on my car?  It's got to close it again afterwards, somehow, right?  And the parachutes have to release somehow as well.

So the drones might use some more of whatever the existing equipment up there is using.

I'm thinking the drone has no pressure tanks.  Cameras, batteries, motors, flywheels, heat shield, and a radio beacon.

[Jim]

1.  How does dragon open that front hatch cover without power?  Is it just a cable like the trunk release on my car?  It's got to close it again afterwards, somehow, right?  And the parachutes have to release somehow as well.

So the drones might use some more of whatever the existing equipment up there is using.

2.  I'm thinking the drone has no pressure tanks.  Cameras, batteries, motors, flywheels, heat shield, and a radio beacon.

1.  Those services are wired specifically for those tasks.  There isn't extra power or signal cables that can be tapped into.  Hence, there is no "existing equipment". 

2.  What is it going to use for propulsion and attitude control (the flywheel along can't do it)

[JamesH65]

Raspberry Pi would have enough oomph to do the math for orientation, could also handle the camera, wifi and you can  plug in an accelerometer and GPS module for location.

Are you talking about just entry or the whole mission.

RPI would be useless since GPS isn't going to help with orientation or location at the moon.  Neither is the accelerometer.

For entry, GPS is not going to help because of the plasma.

At any rate, accelerometer and GPS is not going to provide enough info to point the camera at the capsule.

I thought the thread was about drones? This is a simply throwaway drone. That's all. Something you chuck out of the 'window' to get views of the outside of the craft. Limited capability, limited lifetime, limited cost.

Orientation could also be done with the camera using star tracking, although TBH I was thinking mostly close to Earth stuff where GPS would work.

Just out of interest, if you have two craft in formation, how do they currently know, automatically, where they were in relation to each other, without GPS.

I certainly wasn't proposing anything to video the reentry, I suspect that might be rather difficult.

[IainMcClatchie]

1.  Those services are wired specifically for those tasks.  There isn't extra power or signal cables that can be tapped into.  Hence, there is no "existing equipment". 

2.  What is it going to use for propulsion and attitude control (the flywheel along can't do it)

Okay, so they wire the releases specifically for the drones.  Duh.  It's not like the Dragon going around the moon is going to be exactly the same as a Dragon going to the ISS.  There are going to be a whole bunch of special accomodations.

There is no need for propulsion, aside from a tiny solid rocket motor which drives the drone away from the Dragon.  That might be done pneumatically, I think SpaceX prefers pneumatics.

Attitude control is done with a flywheel.  During reentry, it'll saturate in short order, but we only need it to keep attitude control for a minute or two during reentry, as we're going to be out of range after that amount of time anyway.  Okay, maybe actual flywheels won't produce enough torque, and the drone will need moment control gyros.  But those are more complicated and bigger.

The drone going around the moon can disengage a lot more slowly, maybe even how the Russians deploy satellites from Dnepr, where they let go of the satellite with no impulse and then back the bus away from it.  It will see vanishingly small external torques from radiation pressure, and so should be able to go at least hours before saturating the flywheels.

As for GPS, it would probably be useful to have the drone attempt to get a GPS fix just before reentry, and broadcast that on its beacon, and do the same while it's bobbing in the water.  If the in-space transmission can be received, it'll help the recovery crew find it sooner.

[Jim]

1.  Okay, so they wire the releases specifically for the drones.  Duh.  It's not like the Dragon going around the moon is going to be exactly the same as a Dragon going to the ISS.  There are going to be a whole bunch of special accomodations.

2.  There is no need for propulsion, aside from a tiny solid rocket motor which drives the drone away from the Dragon.  That might be done pneumatically, I think SpaceX prefers pneumatics.

3.  Attitude control is done with a flywheel.  During reentry, it'll saturate in short order, but we only need it to keep attitude control for a minute or two during reentry, as we're going to be out of range after that amount of time anyway.  Okay, maybe actual flywheels won't produce enough torque, and the drone will need moment control gyros.  But those are more complicated and bigger.

The drone going around the moon can disengage a lot more slowly, maybe even how the Russians deploy satellites from Dnepr, where they let go of the satellite with no impulse and then back the bus away from it.  It will see vanishingly small external torques from radiation pressure, and so should be able to go at least hours before saturating the flywheels.

As for GPS, it would probably be useful to have the drone attempt to get a GPS fix just before reentry, and broadcast that on its beacon, and do the same while it's bobbing in the water.  If the in-space transmission can be received, it'll help the recovery crew find it sooner.

1.  Flippant answers like that show that you don't know what is involved.  Adding additional exterior wiring to pressure vessel and heat shield is not simple. There has to be an available pass thru that can accommodate additional conductors.  And no, there doesn't have to be whole bunch of special accommodations for this mission. 

2.  No, they are not going to want a solid motor.

3.  They will just spin the drone as the wheel spin up from zero

The drone around the moon is a great idea and it can come out of the trunk

The entry drone is a non starter

[biosehnsucht]

They are not going to put it on the front.  The whole thread started with putting them in the trunk.  You just don't go and add something willy nilly.  There aren't services like power or data all around the exterior of the Dragon to separate an attached object.  Also there is the safety implications of non separations with the pressure tanks of the drone.

Actually, as far as power and data, shouldn't a normal Dragon have those routed to / through the IDS system for interfacing with ISS? So if you could  remove a chunk of the IDS that isn't needed, to give you the space to put the drones in, you could build their launch platform & etc to fit into the IDS space and interface through those same data/power busses...

That still doesn't help avoid the other issues, and I'm pretty sure during descent they're going to want to close that nose cone up again (since on Crew dragon, it doesn't detatch, but folds to the side), so you can actually eject anything from nose during descent without some major changes to the nose cone to eject THROUGH the nose cone... which is a whole different set of problems.

I think sticking to trunk launched is the best plan, and either giving up on re-entry footage or coming up with another solution.

When the Dragon ejects the trunk, is the dragon pointed prograde or retrograde (or some other direction) ? Because there might be some opportunity to do something from the trunk post separation, but how it's oriented might make things harder/easier.

[IainMcClatchie]

Moon shot:

I agree that trunk launch of a full 4pi steradian camera ball is the best plan.  I further think no radio and no propulsion during flight.  Storage is internal and huge (200 TB), internal IMU and either reaction wheels or control moment gyros to hold attitude.  Data recovery is through internal re-entry vehicle, maybe around 1 kg, with just the flash, a heat shield, and a radio beacon, and no parachutes but direct slam into ocean.

Biggest risk seems to be the huge error for where it might land, as it might land on land and the CEP probably covers a good chunk of the Earth's hemisphere.  I'd probably handle this by using three drones.

Risk of collision with the Dragon would be mitigated by making the ballistic coefficient either significantly greater or less than the Dragon, and then ejecting the drone either forward or backward along the flight path.

Bonus points: send another drone skimming at half or one quarter of the distance to the Moon surface, to get more of a sense of skimming.  Basically send it low enough that there is a little risk of hitting surface features, and then send a few with scatter so that at least one survives.  I'm not sure what other changes to the drone's trajectory would be needed to get it back to Earth... it might need a bit more velocity.

Reentry shot:

This is riskier, less likely, but also very exciting.

Jim, you are correct that I do not know what is involved in spacecraft operation and design, especially manned spacecraft.

Biosehnsucht, great points!  How long before re-entry does the trunk separate from the Dragon?

[philw1776]

Comms during flight is FAR simpler than a re-entry vehicle and recovering it.  Footage is not needed or interesting for over 90% of the week flight.

[IainMcClatchie]

I'd want to target human eye resolution, which is 150 - 200 urad/pixel.  Once you account for some overlap between sensors, the combined resolution of the camera ball will be over 400 megapixels.  At 60 fps, that's 24 gigapixels/second.  The moon flyby will last at least an hour, maybe two.

Good video compression will get that down to 0.1 bits/pixel, so 1-2 terabytes.  Transmitted over the course of 4 days, that's 23-46 Mb/s.  That's a very, very fast link.  Maybe you could use WiFi to transmit from the drone to the Dragon, and let the Dragon bring the data home, but that link would have to stay up for four days over ever-increasing range, or more likely run at very, very high data rates for many hours.  I agree that this is also a reasonable direction to go in, but I don't like it for a couple of reasons.

* Bandwidth constraints lead to downward pressure on quality.  It's best to get rid of these constraints.
* If the drones have to talk to the Dragon, then there are constraints on Dragon attitude, or more likely just blackout periods.  These might be really long.
* Bandwidth constraints will also lead to a resistance to multiple redundant drones.
* Distance constraints will curtail things like super-low skimming drones.

I'm not a huge fan of video compression for high value imagery (e.g. major motion films), so I'd budget 3 bits/pixel for a near lossless coder, or 30-60 terabytes.  I'm not sure how I got 200 TB earlier, but anyway, a lot.  That'll fit in a few cubic inches.  This can be made small.