Adding Automated / Drone Ships

Preamble

I really love the idea of creating ships that aren’t manned by humans or other organics for roleplay purposes and I think it would be cool to have a way to simulate them in game. Here’s my take on how this might work in Aurora.

Automated Ships

Automated ships refer to any vessels which are unmanned by organics. Automated ships function identically to crewed ships with the following exceptions:

  • Automated ships suffer no delay in firing in combat unless it’s maintenance or “morale” (we’ll call it coherence) has been degraded
    • This is referring to the delay in firing that poorly trained organic crews suffer in combat
  • Automated ships carry no crew nor officer. It is instead captained by an AI Core; a sub-sentient computer system that is designed with specific characteristics by their designer. AI cores’ skills are fixed and aren’t improved by experience.
  • Deployment time effects the coherence of the AI core at a slower pace than the morale of an organic crew.
    • However, the longer an auto ship remains at a low coherence, the greater the chance the AI will suffer a malfunction that can cause it go rogue or self-destruct.
  • Effects from hostile jamming are intensified on automated ships, even more so on auto ships piloted by tier 0 cores
  • When automated ships are destroyed in combat, there’s no crew or officers to recover. AI cores either attempt to eject for recovery (risking capture) or go down with the ship to avoid the risk altogether and to provide worse salvage.
  • AI cores add increasing amounts of EM and Thermal signatures to the vessels in question. AI cores can be shielded to mitigate this effect, although this will increase costs significantly.

AI Cores

AI Cores are player designed components that replaces the organic crew. AI cores don’t learn; they are developed, trained, and then deployed.

Automated ships are available even to conventional empires, albeit conventional AI cores are extremely weak compared to crewed ships. The most useful way to use conventional AI cores is as cannon fodder or one-way missions (Probe Droids anyone?). Conventional AI cores are Tier 0.

Tier 1+ cores are player-created components that increase an auto-ships’ capabilities / skills and can be fitted onto auto-ships. AI core’s can be trained to mimic an average crewed vessel’s capabilities or to excel in a specific role / stat line. Overspecializing an AI core comes with diminishing returns.

Some example stats that could be affected by the AI core’s capabilities:

  • Survey Speed (Geo and Grav)
  • Weapon Accuracy
  • Effective Sensor Range(s)
  • Damage Control Rating
  • Maintenance Costs
  • General (Marginally effect all of the above)

Coherenece (Morale)

AI cores aren’t sentient. The only thing they require in upkeep is MSP and specialized coherence testing. Consider how UNSC AI constructs have a lifespan of ~7 years; after which they go rampant. AI cores function similarly in that their lifespan is limited and they need to be reset / re-trained after their designated deployment time aboard a given ship.

Coherence degrades much slower than morale does, but its effects are much more drastic and dangerous. Low coherence AI cores significantly degrade the performance of their vessels until they accidentally destroy themselves or go rogue; at which point control over the AI core is completely lost.

Auto Ships vs Crewed Ships

So, the big question is, what’s the point of having to deal with the crew system if I have automated ships? There needs to be an interesting choice to be had; if one is blatantly better than the other, then there is no choice outside of RP or challenge considerations.

I think the mechanics I laid out previously should provide some interesting choices by themselves, but I will lay it out here.

For Auto Ships

  • Investment into an automated ship is not contingent on needing qualified officers or crew, only materials to create them
  • Automated ships are more suited to longer deployment times
  • Automated ships can be further specialized at the design level with greater efficiency: without the need to carry crew, automated ships are more compact and cheaper than an equivalent organic one
  • Sustaining combat damage doesn’t result in the loss of crew or officers
  • Automated ships can be designed to make defeat provide less material / intelligence for the enemy from combat losses
  • Autonomous ships can take the place of ships that may have previously been piloted by conscripts

For Crewed Ships

  • Doesn’t suffer from additional ECM or jamming effects nor do they have an increased sensor profile
  • Officers and crew are “free” (not factoring in the cost of academies) and don’t require you to invest minerals into making them better, just training and time
  • Crewed ships can get more effective over time as a ship’s crew gains experience
  • Recovering skilled officers or crew can preserve valuable talent, even between new generations of ship designs
  • No additional maintenace costs or complications from needing to maintain AI cores
  • Low morale ships don’t go rogue
  • Less research and development time overhead compared to auto-ships.
  • Crewed ships benefit from well designed and staffed naval organizations whereas automated ships do not.
  • Allowing organics to captain your ships centers your heroic naval commanders and their achievements. You probably don’t want to decorate your CPU with honors.

Misc

Some additional balancing things that may need to be considered

  • Crew quarters are replaced by an AI core equivalent that is lighter and cheaper and would scale similarly to crew quarters so that deployment time is cheaper to buy but not free.
  • 6 was an arbitrary number for the AI core tiers. Whichever amount is the most straightforward and interesting to implement is the best.
  • The size of AI cores could be tied to some sort of multiplier for the amount of its effects similar to Fire Control. You could plug in a massive 2000kt super computer or a 50kt calculator depending on what you want.
  • At high enough tech levels, perhaps you could research a “sentient” AI core that doesn’t start out with any proficiencies but can learn and improve a ship in the way that an organic crew does.
  • I have no idea what minerals AI cores should take to create.
    • Tying back into the idea of a size system like the ones engines / fire controls / sensors have, the player should be able to choose between making super-advanced and effective AI core models VS super-expendable and stupid models.
  • Anything else I haven’t thought of or oversights on my part
1 Like

One of the reasons I do support the idea of AI ships is particular spoiler - precursor automated guardians - with it’s at least thousands of years of existance, and unfixable instability of AI Cores contradicts this storyline. Therefore, I think the deployment time (coherence loss rate) needs to be a developing thing. It’s ok to make conventional cores losing coherence in 5 to 10 years, yet higher levels have to grow in stability at a very steep rate - something like 10 times a level, to make usual L4 spoilers believably stable for thousands of years.
What I’d suggest to implement instead of catastrofic instability is incremental degradation. This would just add consistency.
For the same reason, I’d suggest to lower the maintenance cost for AI ships, or (even better) make it decreasing steeply with tech levels too.

Specialization capability makes AI too strong, I’m against it. We’d just need too much of (disbelievable) flaws to couneter this capability.

Increasing sensor ranges adds significant complexity to the code and planning, and I think it’s not worth the interesting choices it brings. Just consider the circles on the tactical map - they’ll either went wild rainbow trying to represent different crew/core influences for a present set of sensor models, or became misinformative without representing it. Aurora simplistic user interface just badly suited to mitigate the complexities of this kind of mechanics. I think it’s better to let sensor and FC ranges as much limited in variations, as it’s possible for the general Aurora design.

I don’t like fixed AI vulnerability to EW, it’s not natural. Sensors is one thing and core is completely another one, it has to be isolated from non-penetrating effects. Adding a risk of hack with significant tech disadvantage is better, I think. Small risk, yet it’s a small risk of very significant accident. Adding specific strong vulnerability to meson weapons is also a natural solution (call it “meson AI core shock” or something).

50kt calculator isn’t any kind of believable anymore, sorry. We cannot un-expierience the last 50 years of microelectronics development. It’s better to fix AI Core in some small size, yet just limit it’s capabilities.

What else I think is worth adding to the AI ships flaws is prolonged jump shock. It’s a bit counterintuitive, the first thought we tend to have about automation is that it’s more suited for any kinds of shocks. Yet jump mechanics is complete SciFi and not very hardcore one, it doesn’t add much disbilieve actually to specify that jumps do destabilize electronics much more then it’s much slower and generally less reliable biological alternative.

Also, as I have suggested already - AI ships may be much less or even negatively assuring for the biological population (desreased, zeroed or even negative Police Value). Totally consistent with how we do react on actuall combat AI.

3 Likes

As to the PPV you mentioned, hands up who wants armed skynet orbiting their planet?

Sarah Connor: Boy, That's A Tough One

2 Likes

Personally I would be interested in drone-ships specifically in the context of drone fighters. It’s what makes the most sense – they’re controlled from a carrier or tender rather than semi-autonomous, the primary goal is removing live crew from a very high-risk role that gains limited benefits from training/officer command, and they benefit heavily from any incremental increase in volume efficiency.

Drone warships that aren’t just flatly overpowered on the upkeep side of things create tonal dissonance re: certain spoilers as @serger noted. But if they’re not flatly overpowered it’s an extra layer of complexity to half a dozen game systems which are already very dense for little good reason.

Implementing them on a broader scale IMO should be a race design modifier that affects base rates for everything an empire does rather than one more checkbox for ship design. That would also translate cleanly into implementing some other SF/space opera tropes like hivemind empires, clone empires, and post-/trans-human empires with a high percentage of cyborgs, uploads, &c.

1 Like

I’ll say this much - automated/drone options would be nice from a purely RP point of view. I feel so bad for all the “volunteers” on all the light scout shuttles that’ve died a horrible death jumping into NPR jump-point camps over the years. :grinning_face_with_smiling_eyes:

1 Like

“Optional crew space for an engineer to maintain the drone” is a load-bearing phrase in my RP setting.

4 Likes

Yea, and then you have to un-see every False News message about these engineers dying.

3 Likes

Thanks for the comprehensive reply! You’ve given me a lot to consider.

  • Coherence / Instability
    • I actually like your idea better. Having the robot ships start out only being suited to deployments a max amount and being able to research longer deployment times makes a lot of sense. Also introduces another level of investment into AI ships that makes the choice more interesting.
  • Specializations
    • Instead of specializations, perhaps the stats of the AI core would be something like this:
      • Coherence Resilience / Longevity / Durability: How long the AI core can deploy without suffering additional coherence loss penalties.
      • AI Core Capacity: Bonus to AI skills at only a small increase in cost.
      • AI Core Overclock: Squeeze out additional performance for diminishing returns. This simply increases the crew grade / training and adds a generic tactical bonus similar to an officer.
        • This efficiency should be inferior to bonuses that you can obtain from an organic officer (especially ones with stacked naval organizations) to keep AI ships leaned into the idea of being cheaper and more expendable whilst still creating a design choice for the players
    • This would negate the potential for overpowered or weird builds with AI ships while having more basis in the way that skilled captains are represented in Aurora as of now.
  • Electronic Warfare / Sensors
    • My justification was more that an AI core would be less adaptive or “imaginative” and therefore more susceptible to jamming (and to introduce another level of being weaker than organic ships). If we feel additional maluses for AI ships is unnecessary, then this is an idea that should be discarded. Not to mention the map rings…honestly you have a point about that. In my game the map already feels cluttered with the amount of different sensors that are showing up.
  • AI Core Size + Cost
    • The size of the AI cores was more to keep the tech in line with something like fire controls as the effectiveness increases both cost and size. If we feel that is too absurd, perhaps we could keep it to simply increasing the cost with little to no effect on the size of the AI core.
      • The size of the AI core could represent the AI core itself in addition to the hardware needed to link it to the ships systems as well as the rest of the machine / room that houses it.
      • Also I was being mean to our hypothetical robot friends by calling it a “calculator” at 50 tons :grinning_face_with_smiling_eyes:
  • Maintenance Costs
    • Not much more for me to add here, although I agree that the primary appeal of automated ships should be trading less effective ship piloting for reduced maintenance, playing into the idea of these being high-endurance vessels over all else.

Also I really like your jump shock, meson weapon, and PPV ideas. If we do get robot ships I’d like those to be implemented.

Actually I’d like to have at least Tactical AI bonus (which is currently the only one really valuable) easier to boost, comparing to training officers. I have seen modern combat AI, and while some media greatly overestimate it’s prevalence and influence on combat activities here, it’s still very impressive as a tool for long range strikes in heavily jammed environment with a lot of masking, false and civillian targets. I’m sure it’s quite easy to train AI to be much better then any human being in this particular task of finding valuable targets, especially when functioning in so inhuman environment as open/transnewtonian space.

So, the primary AI ship malus I’d like to see is a risk of several interconnected catastrophic events:

  1. Population run amok against them in large scale Butlerian Jihad or at least instant secession when AI fleet is locally stronger then crewed one (people would demand switching off and scrapping AI combatants).
  2. Quite easy to kill or hack with mesons, any meson-heavy opponent makes AI battle fleet too much vulnerable.
  3. Small, yet unavoidable risk of getting out of control (becoming hostile) without any visible reason.

And would these parameters be set in database (not hardcoded), I’d be happy to adjust them as I like for my campaigns and then never try to convince anyone here to support my suggestions about it. :slightly_smiling_face:

Honestly id be happy with just a checkbox that replaces the crew spaces with something like automation compartments, removes the stat showing crew count and stops a lifepod from launching when it’s destroyed.

I’ve been happily playing as if automated ships are a thing already and just ignoring the body count, so would be perfectly happy if they worked exactly as crewed ships do but I could send them on one way missions with a clear conscience.

3 Likes

Completely valid take. For now I just slam the mute button on my comms device when I hear word of “survivors” in “lifepods” from a destroyed drone fighter…

Well said. Now that I think of it, “drone launcher bay” or “probe bay” is a great use-case for the miscellaneous components feature, isn’t it?

1 Like