Robot ships

[esampson]

. . .About the price, as I am talking about fighters, I guess the standarized design could well apply, as it's unlikely only one or two are buit, but large quatities of them. And even if it does not, as the use of the robot brain saves you equiping it with a computer, and that's a major saving...

I didn't realize you were talking about fighters. Yeah, there could be a lot of good arguments made for robot fighters that wouldn't exist for robot starships (the space saving on a starship wouldn't be that great when expressed as a percentage of the ship's body, but for a fighter they could be quite significant, which would make it a smaller/more agile target).

Also fighters don't operate independently for long periods of time and their theater of operations is more restricted (which makes things easier for the FLP).

 

[aramis]

So what is the difference between a Robot Pilot with 'Data', 'Autonomy' and 'AI' logic? All three have 'pilot skill'.

And then this previous question resurfaces:

These are all normal decisions made by a human pilot that have nothing to do with the skill to actually operate a craft.

Data: provide list of targets and specific status before next target. Logic needs objective measures. Follows directives pretty exactly, "Return hostile fire" may result in shot rock-throwing monkeys, nut dropping corvids, and small animal dropping eagles; but won't shoot at enemies who are unarmed or not shooting. "Mow the General's Lawn" may result in mowing the flowers, too. May not be the right General's lawn.

Autonomy: provide mission objectives and secondary objectives. Can be slightly fuzzy ("Report and engage any know hostiles" or "Secure the base perimeter. Allow access to persons in two-three-zed-niner list."). "Return Hostile Fire" probably won't shoot the critters, and may challenge known enemies, and will take precautions against armed enemies. "Mow the general's lawn" won't get bedded flowers, but may get any not in clear beds; may opt not to mow until the generals kids are not playing on the lawn. If there are two generals, may ask for clarification, or may just do both, as time allows. Depending upon gardener skill, may or may not also rake and fertilize the lawn.


Low AI: treat it like an Airman Basic, seaman recruit, or Private E1... If you tell it what needs to be done, it will infer the correct sequencing, and may make reasonable choices as to timing. "Mow the General's lawn" will be the general in their chain of Command. If multiple, will infer from "The General's" to be the senior most. Won't mow if children present unless time requires, and will ask children to move if need be. May put toys on porch, but unlikely to mow them. Won't mow if it's raining and the mower is electric.

High AI: Decides not to mow the lawn because it's busy listening to the general's wife whinge.

 

[mike wightman]

These are all normal decisions made by a human pilot that have nothing to do with the skill to actually operate a craft.

Yup.

Who tells a military helicopter where to fly next?

The crew don't decdie - they follow orders.

 

[atpollard]

Yup.

Who tells a military helicopter where to fly next?

The crew don't decide - they follow orders.

Perhaps I misunderstood ... I thought that you were among the group advocating for an all robot crew civilian starship ... kicking that decision to the "Low Data Logic Robot Captain" doesn't really help. Letting the human captain/owner order his robot pilot around makes perfect sense to me (although for a pilot, I would prefer Low Autonomous Logic" as a minimum).

 

[mike wightman]

Not an all robot crew, no. But certainly the ship could be a robot.

Human military personnel may as well be robots - they follow orders. Megacorp ship crews follow orders.

You could not have a robot free trader, but you could have a robot crewed free trader ordered around by a human owner.

Similarly you could have military ships that are almost entirely robotic but with humans ordering the robots to do stuff - think Andromeda.

Artificial sentience - the Traveller TL17+ robots - make all the decisions themselves and remove the need for human direction. You could ask such robots to work for you , but they could turn around and say no. Or the robots could decide to make their own empire

Come to think of it, grunts could say no, but they rarely do

So going back to the original post I can easily see the Hive federation or some solomani using robotic freighters and x boats on safe routes. More experienced robots may be used on more hazardous routes. For uncertain encounters the solomani probably wouldn't trust robots at all and would send mostly human crews (with robot helpers since they started this), whereas the Hivers trust their high TL experienced robots to be diplomats, fleet commanders, even entire armies.

 

[esampson]

. . .Similarly you could have military ships that are almost entirely robotic but with humans ordering the robots to do stuff - think Andromeda. . .

The only caveat I would have to that is that one robot ordering another robot probably wouldn't be much better than a robot making a decision wholly on its own (as the ordering robot would not realize that the working robot was making a mistake that needed to be countermanded) so in the case of ships where you have things such as dozens of engineers working per shift you would probably still require multiple humans to oversee them.
Your basic limit would be you would need enough people that they could give each robot an instruction. The higher the FLP the less oversight each robot would need and the more robots a single person could control.

You could probably turn it into some kind of easy formula (e.g. 5 Low/High Data per person, 10 Low Autonomous, 20 high Autonomous).

 

[mike wightman]

I quite like that

But again, robots learn, so perhaps less human oversight is needed for higher skilled, more experienced robots?

I fully agree that until TL17+ robots can't decide their own future, but within their skill set they can determine how they cope with the present. They can't decide to just do stuff - a robot plugged into a 200t free trader can't just decide to go off speculating - but it could be commanded to jump from system A to B to C an back again and pick up goods from the broker.

 

[esampson]

I quite like that

But again, robots learn, so perhaps less human oversight is needed for higher skilled, more experienced robots.

I fully agree that until TL17+ robots can't decide their own future, but within their skill set they can determine how they cope with the present. They can't decide to just do stuff - a robot plugged into a 200t free trader can't just decide to go off speculating - but it could be commanded to jump from system A to B to C an back again and pick up goods from the broker.

Well, you could probably do all sorts of things with formulas depending on how many headaches you are willing to put up with.

A person who is controlling more robots than someone else has less oversight, so the odds of one of their robots making a poor choice increases. Robots with higher skills might be less likely to make a decision mistake than a comparable robot with the same FLP. Older robots (excluding Low Data) would be less likely to make a bad decision than newer robots. Etc.

 

[McPerth]

One thing to discuss too about robot skills is if they also give the robot the additional knowledges a skill gives a player, specifically:

• In MT (and to lesser extent CT, as they are rare there) the included or serves as skills
• In T4 or MgT the cascade skills
• While in TNE there are also cascade skills, I guess not many robots are used there, as the phobia to any AI is well justified there, except for the vampire ships/fleets, but I guess they're excluded from this discussion...
• I don't know other versions enough to talk about them

While for a sophont (or at least to a human) is quite difficult to learn some skills without learning some basics of others, a computer (or robot brain) could be programed to perform several functions without knowing about those basics.

This was shortly discussed for MgT and here's the conclusion I reached (or course as argueable as any other):

Just to clarify:
The Expert software gives its rating minus one as skill level to a character or a +1 DM (not -1 DM). With Intellect software the computer is treated like a character regarding Expert software.

So a Repair Robot per pg 95 with Intellect/1 and Expert Engineer(any)/2 does skill checks as Engineer(any)-1. And Engineer is a cascade skill.

While this is how cascade skills are handled, I'm not so sure expert software also gives you other skills in the same cascade group at 0. I know this is not explicited in any way in the Core Book, and so regular rules should be used (as you did), but, using common sense, and IMHO, to allow you an expert program than gides to performing some tasks you don't need the theoretical/practical knowledge that you need to truly have the skill yourself, so I'd see some logic to limit skills given by Expert software to the specific skill, without any effect to other related (cascade) skills.

I know this discusion is a little old, but I have been thinking about it since then, and now, while reviewing the rules, I think things may be clearer than we thought (albeit I agree this is only one possible interpretation of them).

The key is that rules (page 92 of the Core Book, software table, under expert programs) state (as you quoted above) The Expert software gives its rating minus one as skill level to a character or a +1 DM (not -1 DM). With Intellect software the computer is treated like a character regarding Expert software. (emphasis are mine).

If you put on a computer (or robot, in this case) an expert software for a cascade skill (Engineering in this case) it will serve also as a 0 rated expert system for all related skills in the same cascade. As the rules quoted above tell about rating minus one as skill level (not rating as skill level with a -1 modifier), the skill is given (the robot has it in this case) at one level below 0, and that means it has not the skill.

As said, this is only one interpretation of the rules, and as such, open to debate, but IMHO it may as well resolve the case.

 

[Carlobrand]

"Robopilot" capability is probably built into the existing ship's computer since the ship only requires the one pilot for 24-hour operations. It lets the pilot get some sleep, go get some dinner, go socialize with the passengers, and it chirps an alert if it encounters anything outside of the routine. In that guise it doesn't need to be terribly bright - just recognize when something's within sensor range or when the ship's not responding to controls properly.

Occurs to me that bots could be quite useful as ship's gunners, although that moves beyond Imperial tastes. In the gunner role they can be pretty tightly supervised - weapon's only hot when the bridge releases it for combat, bridge can designate friendlies and unfriendlies, the gunner bot only needs to operate the gun with respect to the assigned target.

Given the robotic preferences of the Zho, it's possible they're already doing that.