What would be the best design for a ship maintenance bot?

Question

Yeah, I know I asked a question about a medbot just the other day but my book has multiple AI "characters" [dunno if they count as characters] and one is a maintenance bot [at least it used to be]. For reference the setting is a high-tech space-faring society, though it is still pretty hazardous – even short colony-to-colony trips are still dangerous and take months, and a lot of people die from catastrophic failures from faulty/cheap tech; damage to [space]ships is common since there's a lot of unrest.
Maintenance bots, in this world, are used to crawl around the inner the ship [generally beneath the floor in a so-called "maintenance space", very creative], and makes sure everything's still working. "Everything" being hydroponics systems, water/air filters, the engine, some electronics, door mechanisms, and pretty much anything else you can think of. Keep in mind they do alert humans of these errors, and will let them take over the job if necessary. They're their in case of an immediate emergency that needs immediate attention, and a human can't get there easily/it would be dangerous for a human to do.
Basically I'm trying to design a maintenance bot primarily off function/practicality, and since I'm bad at designing robots [and you gave me amazing answers last time], I'm asking you again :)
Also, if you need any sort of clarification/have any sort of critique [as long as it's reasonable], please tell me!!

Answer 1

Multiple locomotive arms with magnets and rollercoaster style runners built into the ‘claws’, then multiple utility arms capable of switching in/out a variety of toolsets all built around an internal bay that can contain replacement parts (or a 3d printer) and the various tools required for repairs. Attach various sensor/communication arrays and diagnostic interfaces at the nominal ‘front’ and you’ve got a fearsome looking design bristling with usefulness.

The locomotive arms can either crawl by attaching one at a time to points inside/outside the hull or, if they’re available, lock onto dedicated guide rails and engage some electric motors to get around the ship fast. If it needs to switch rails or move over a damaged portion of ship then a second arm can be engaged before the first releases, making sure it never loses contact with the ship. The sensors and diagnostic interfaces spot problems that the ships sensors haven’t, the communication arrays network all the repair bots and the ship.

Then if any damage is detected a repair bit will either repair it on the spot with it’s utility arms and 3d printed parts, or will summon one of it’s compatriots with suitable replacement parts. If the damage is minor repairs can be effected at blinding speed thanks to the locomotive guide rails and multiple utility arms. If it’s major repairs can still be completed because the bot can crawl independently, even in 0 g, and multiple bots working in concert can move even large pieces of equipment across the ship.

For added security you can add a couple of magnetic grapple guns and some limited gas thrust capability, making your bots the nightmare spawn of a shrimp, a spider, a squid and a tool shop.

ADDENDUM:

You might also want to consider making your AI a swarm consciousness. Having dozens or hundreds of highly networked repair drones in a single, coordinated system makes much more sense than having multiple potentially conflicting AIs.

Plus you get a swarm of shrimpspidersquidtoolshop bots instead of just one!

Answer 2

Cockroach.

This form is ideal for scuttling about, squeezing through very small crevices, walking on walls and even ceilings. Capable of operating in a wide range of lighting conditions, including complete darkness. Has advanced tactile and olfactory and chemical sensors for environmental awareness.

You would need various sizes, for the various specialities. Tiny ones for electronic maintenance, medium ones for garbage removal, large ones for structural work.

It should be easy enough to adapt the low, many-legged form for outside operations, possibly with magnetic or electrostatic contact pads on the feet.

You do not need to design your bots, nature has already done the field work for you. All you need to do is to figure out the most effective way to manufacture/program/train them.

You might even want to imitate not just the form, but actually adapt the biological organism itself. With suitable controls, you have a self-replacing robot workforce that can get anywhere.

For a fuller answer, just glue Joe Bloggs' answer directly to this one, the sum of the two is the full answer.

Answer 3

The name of the game for these robots is prevention - within the available "maintenance" space, there will be too little chances for actual repairs. There may be some work in emergency patching, but I would not call this "maintenance".

In brief, I propose you:

• robots acting as mobile sensors platform, scan for troubles in advance and pinpoint the detected weaknesses.
  Use X-ray and ultrasound for metal pipeworks, ultrasound only for plastic pipeworks, sensible thermal camera for electricity, wall/floor vibration sensors for things that rattle when they shouldn't, even amplified vision for inspecting vibrations ^{(watch this linked YT, highly recommended)}
  Since only the platform needs to be mobile, you can use interchangeable sensor payloads.
  
  A static network on sensors won't be as thorough, because there will be places in which powering and connecting those static sensors to comms would stay in the way of the main utilities.

• self-preservation AI, running on a different neural system than the flight (and weaponry?) control systems. Pretty much as the immune system and the enteric nervous system doesn't run on the brain.
  The static and mobile sensors will feed into this subsystem which may compile a report to the humans in terms of "arthritic pains in attitude control" or "slightly feverish air regeneration unit" or "food poisoning in the left engine"
  Of course, when asked what are the real causes and what needs to be done, the AI will switch to a more technical language and possible present a set of plans for remedial (set that may be empty, in which case the diagnosis may come "Its dead, Jim, now you're properly effed")

Note: the details about robots mobility and their number is considered a trivial problem: whatever fits the bill.

Answer 4

It depends on some certain factors.

The biggest one is if your world has artificial gravity on the spacecraft or not. If not, then you'd find that it would have reaction control thrusters to maneuver quickly through the weightless environment. If not, it would have legs to scamper around in the space. Think a mini version of Boston Dynamics's Spot perhaps. Maybe it would also have adhesive footpads and such.

A big thing is that it would have robotic arms for doing the various tasks needed, maybe multiple if it needs to do things quick and doesn't want to have to take the time to pull in the arm and change out a tool.

It'd probably have a camera on it to give the operator a view of what was being fixed, to make sure the robot is doing its job.

In reality, a robot like this is not too far away - Spot is rather close to this - you just have to figure out how to make it climb on ceilings.

Answer 5

I would think something like a squid (with any number of tentacles) would work for the majority of applications, and it would be a soft body robot, rather than a typical hard plastic body.

The soft body approach would allow the bot to get into tight locations as well as form around parts and tools for manipulation and brace themselves for stability. Some parts of the robot could still be hard, to help reinforce a stance for stability when handling heavy pieces or during hard maneuvering, but that would be a bare minimum of rigid pieces to perform those specific functions.

The robot could house typical tools in it's body crevices in the same way a carpenter has a leather toolbelt. There could be specialist versions of this bot, so if something requires a special set of tools, it can be called as backup or replacement for the standard bot that went out to investigate.

It doesn't matter that parts of the ship aren't pressurized, since the movement would still be controlled in a similar manner as how it works in a pressurized area. In a habitable space, the robot moves by "air" being pushed into the sacs and then sucked out. In vacuum, the "air" would be drawn into the sacs and then sucked out. This gas doesn't have to be air, but rather nitrogen or any easily sourced gas that minimally reacts chemically with components, such as Noble gasses.

As the bot is on a ship, the skin would have to be tough, since it will it move past sharp corners, general hardware, general friction, and it's own sometimes sharp tools. This means it's already likely going to be able to handle the pressure differential between the vacuum of space and it's own sacs. Sudden decompression of a compartment/room/hangar/whatever would pose a threat, but as long as the skin remained intact, it would only incapacitate the bot for a very short time while it recalibrates the necessary pressure in each sac to remain working. Depending on how slow a decompression is, the robot may not even be incapacitated, as it would have time to adjust as the pressure falls. It would likely have more concerns of being sucked out with the air than continuing to do delicate work.

With this being a flexible robot, it can be designed with flexible circuit boards. Some of the circuits might even be incorporated into the body and tentacles, itself, allowing for a smaller volume of space that ends up not being flexible in order to store the pumps/motors and any other components that have to stay rigid.

There will likely be fewer motors/servos/actuators/pumps and them being centrally located, it'll be easier and lighter to shield them from various forms of radiation. And because the electronics for this bot will likely be simplified and the majority of higher order functions being completed remotely to reduce circuit size, the shielding will be considerably less, too. The actuators might even be as simple as solenoids.

Then again, some of these parts may be flexible, but less than the rest of the squid, due to synthetic muscles. Using these synthetic muscles, there may not even need to be "air" sacs, but rather these be used in the tentacles themselves. This would avoid the need for a gas reservoir to be carried around. There are even flexible batteries and flexible solar panels to power the robot. Given enough development, these robots could potentially be only a few millimeters thick for storage, minus the tools.

I was thinking that suction cups could be used on the tentacles, but that relies on air pressure around the suction cup to hold the cup when suction is applied. This works well in an atmosphere, but not in space. Instead, small electromagnets could be used instead. They could be spaced out and individually activated, just like suction cups. These would be embedded into the skin, to prevent damaging the electromagnet and whatever it's attaching to from bumps and scratches. Suction cups could be an alternative on the tentacle, when moving in a space that's non-ferrous. Other forms of attachment can be used when on carpet, liquid, or pretty much any other surface. With this being a flexible bot in no/low gravity periods, it could even use jets of air or waves of a flattened tentacle to "swim" through a pressurized area.

Answer 6

I would say it depends on the ship. I'm rather fond of the DRDs from farscape. Little wheeled pods, with independant eyestalks

While there's a few 'special' ones, the main advantage there is they're a swarm of little robots, able to wander around the ducts to check and fix things. I would use these as the 'core' of a more versatile systems with little diagnostic repair drones being able to control bigger 'dumb' chassis as needed.

While like ants, I guess a large number of these could do bigger tasks, you can't exactly have a dozen of little shoebox sized drones balanced on each other. For some reason the 'wheeled pod' form factor is popular. Amusingly the local police trialed the two form factors I think would be perfect here

(source)

One large wheeled machine used to move around parts and equipment, one large, 'humanform' machine, though with arms for tasks that need humans. In theory you could use a boston dynamics style 'big' dog bot - as per this Tom Scott video.

Finally you need some way to do space based inspection and repair. While 'organic' space flight capabilities would be nice - its weight you don't need.

Babylon 5 had little repair pods based off the starfuries

Basically a cockpit, engines, manipulators and just enough structures to hold it together.

I'd pair it up with a smaller 'thruster pod' - maybe something inspired by the multiple kill vehicle - essentially multiple jet thrusters allowing for ludicrous agility.

Answer 7

The top half of a humanoid.

https://www.stanwinstonschool.com/blog/aliens-movie-alien-queen-attacks-android-bishop

In Aliens the part where Bishop comes back is the best. I cheer every time. What, you have not watched Aliens? Callow youth! But you are in for a treat. Go watch it now!

Ok, you're back. Bishop was a half android by accident which is why he is so messy in this picture. But Maints is half an android out of practicality. You had a lot of humanoid androids and a lot of the ship was built so that humans could access it. The top half has all the needed parts and it is more nimble without the lower half.

Sometimes Maints uses a wheeled dolly getting around populated areas of the ship. Just as often it walks bipedally on its hands. It is lightning fast on ladders or webbing. All that gut stuff is tucked in of course. It is still more than a little creepy, especially when it emerges unexpectedly from a vent.

Maints still has its legs around somewhere. It can get them back on if the princess visits or there is some other formal occasion.

Answer 8

If it's largely an inspection bot it will be checking various components/areas for temperature, vibration, pressure, corrosion and leaks either chemical or radioactive. If it has limited space to move through it should be small, or at least have a small cross-section. For creepieness I'd suggest something snake-like with a flexible body that can charge sections of its skin electrically or magnetically so as to achieve cohesion there and using a slithering motion to use those sparse sections of cohesion to move freely among bundles of piping, cabling and through small crevices in structural work irrespective of gravity. It would have sensors in its skin to measure the factors above and rely on direct contact, by slithering over equipment or through spaces, to take its measurements. For added danger I'd give the snake-bot a behavior of constricting structural members and slightly compressing them, linearly not axially, to check that their stress/strain curve is linear. This is important for determining if structural loads are within design specification but also gives the snake-bot sufficient strength to be dangerous.

Answer 9

Your ship has to be designed for members of it sentient species to do maintenance as well. That means that there are maintenance spaces and access ways.

They need this in case the droids are taken offline.

So for a human spaceship, a human shape is best.

Generic universal answer is a sapient shaped droid made of a large aggregate of nanobots, so it can change to whatever shape is needed.