Ballard Robotics Brains

[samuelvss]

Semi-autonomous brains: LBB8 DESIGNS

Ballard 2AAA BRAIN TL13 MCr. 2.67 vol: 181.8 liters weight: 37.8 kg
CPU: 70 parallel, 8 synaptic. Storage: 280 standard, 60 synaptic
Runs full command, High Auto Logic, and 28 application programs
Apparent Intel 13, Educ 13 (13)

This is a the leading brain made by Ballard Robotics, and it exemplifies the capacity of the AstroMech chassis, making a droid that can take over any function of any crew member of a small starship requiring shipboard skills. Significant ability to learn.

Ballard 2A BRAIN TL13 MCr. 1.6 vol: 128.6 liters weight: 27.1 kg
CPU: 50 parallel, 6 synaptic. Storage: 200 standard, 30 synaptic
Runs full command, High Auto Logic, and 20 application programs
Apparent Intel 13, Educ 13 (13)

This is the highest brain normally seen in semi-autonomous robots. In the AstroMech, it allows for a droid that can be tailored to do two or three jobs well, including piloting small craft in combat. Some ability to learn.

Ballard 2D BRAIN TL 13 MCr. 0.94 vol: 145.6 liters weight: 21.6 kg
CPU: 50 parallel, 6 synaptic. Storage: 160 standard.
Runs full command, High Auto Logic, and 16 application programs
Apparent Intel 13, Educ 13

This is the lowest brain Ballard makes for semi-autonomous robots. It is incapable of learning, but can replace any crewman under normal conditions, and two or three in a pinch.


Non-autonomous brains:

Ballard 1C BRAIN TL13 kCr. 185 vol: 68.1 liters weight: 14.1 kg
CPU: 10 linear 10 parallel, 1 synaptic. Storage: 120 standard
Runs full command, Low Auto Logic, and 12 application programs
Apparent Intel: 4 Educ: 12

Ballard 1D BRAIN TL13 kCr. 175 vol: 47.1 liters weight: 10.1 kg
CPU: 10 linear 10 parallel, 1 synaptic. Storage: 80 standard
Runs full command, Low Auto Logic, and 8 application programs
Apparent Intel: 4 Educ: 8

This is the lowest brain Ballard makes for AstroMechs. It is ideal for a “slave,” that can also perform basic starship functions with competence in a pinch. Fine for a gunner droid with basic damage control and rescue functions.

Ballard Robotics is a wholly-owned subsidiary of Ballard Designs, LIC, currently headquartered at Kandom, 2116, Diaspora. It was originally a Solomani company, but has severed ties with its assets in Solomani space in 1103, after the Solomani Rim War. There are several local naval architecture, shipbuilding and robotics companies throughout the Solomani Confederation that are the remnants of Ballard Designs in Solomani territory.

Ballard Robotics has been the leader in making robots that are integrated with starships, specifically those of Ballard Designs, but their robots are made to be generally competent aboard ship. Each human workstation has a AstroMech workstation or interface co-located with it. The AstroMechs are made with the intention of replacing one or more crew members. The 2 series are made to be “semi-autonomous,” but really only human customs and imperial regulations keep a couple from running a starship by themselves. Ballard ships are often shy on staterooms for gunners, medics, engineers and small craft pilots because it is cheaper to take the MCr .5, and put a down-payment on a 2 series. The 1 series really do require more specific directions on what tasks to perform, but are perfectly capable of performing them. A 1-series gunner will typically stay in the turret, and await commands. The apparent intelligences are important in this.

 

[atpollard]

Robot brains for auto-pilots is something that I found missing, but useful in Vehicle/Starship design. Good work.

Two questions:
1. Is this T20 rules? (Int 18 means one thing in T20 and something very different in Classic Traveller) I assume that it is T20.

2. What TL are they? It would be nice to see the TL listed in the description rather than needing to look up the components in the book.

 

[samuelvss]

All are TL13. These are LBB 8 designs; the intelligence, and other stats, are therefore CT. So yes, the 2AAA is smart...as close to AI as we're allowed to get before TL16, I believe.

The second number for education is, for the 2AAA and the 2A, the maximum level that the brain can be taught up to.

They are actually meant to go into an R2D2-esque AstroMech. An autopilot is a good thing, but it won't drag your unconscious butt into the airlock...

The chassis, with a fair mix of sensors, powerplant, and batteries, costs about kCr160, so the real cost is the brain. That being said, the brain could instead be mounted directly into the equipment it is to "control."

I like the idea of robot gunners, for starships and small craft, robot medics, and robot astrogators for scout ships.

I did these off a LBB 8 spreadsheet, which I have since lost. There is a link to it in Freelance Traveller, but it has gone dead.

 

[atpollard]

The LBB8 rules (or the spreadsheet) seem out of balance with the rest of the game.

An intelligence of 12 represents the top 3% of the population - that corresponds to Gifted and Mensa levels of intelligence. Int 15 represents Human Maximum - Newton/Hawkings level of intelligence. The TL 13 robot at Int 18 is as far above 'Newton' as 'Newton' is above the graduates of a typical college.

... And full AI does not occur until TL 16?

Strictly IMTU personal opinion, but I would view AI as Int 7+ (as smart or smarter than the average person) and pre TL 16 as restricted to Int 0-6. If we set maximum Robotic Intelligence at TL-9, then Int 7 (human average) would be obtained at TL 16 (the traditional start of AI) and Int 2 (the bottom 3% of the population) would be obtained at TL 11.

This is not a critcism of your work. I applaud the concept of simple 'plug and play' assemblies built up from complex rules and I assume that your designs comply with LBB8. It is the LBB8 rules compared to the earlier Traveller assumptions that I have a problem with and I am critical of.

 

[BeRKA]

R2-d2

With the rules from book 8 You can also teach the robots skills. Quite high level of skills... And this means that You can have robots as expert gunners or fighter pilots... :devil:

But this would of course mess up the strict starship automation rules... :nonono:

 

[samuelvss]

The LBB8 rules (or the spreadsheet) seem out of balance with the rest of the game.

An intelligence of 12 represents the top 3% of the population - that corresponds to Gifted and Mensa levels of intelligence. Int 15 represents Human Maximum - Newton/Hawkings level of intelligence. The TL 13 robot at Int 18 is as far above 'Newton' as 'Newton' is above the graduates of a typical college.

I do not disagree. For what I can pay for a Model/1, that can't fly the ship, I can have a trashcan that can fly the ship better than 98% of pilots...

It can unbalance the game, but is "canon." The key is how, and why, the GM puts this into the mix.

Note that the key here is "apparent intelligence." A 2AAA, with a superhuman "apparent intelligence," is still only semi-autonomous. There are things, by way of human nature, that can not be programmed.

By way of analogy, my [crappy little] computer is a better speller and mathmatician than I am, by orders of magnitude. I could program a computer to ace an IQ test. I can create an "expert system," that reacts tactically. By some measures of IQ then, my computer with the right software is already smarter than me. It, of course, could not write this email. As TL's increase and more automation is sought in the drive towards AI, more areas in which intelleigence can be measured will be filled up.

The Ballard 2AAA would not be able to decide whether or not it was eithical to cause the young medic to fall in love with it, let alone effectuate that choice (with or without the amazingly skillful use of appropriate prosthetic packages!) decide whether writing a poem was appropriate, or even make up an appropriately silly song. The 2AAA could, however, do what it is programmed to do better than a human, possibly an order of magnitude better. That programming will be, however, in certain specific areas, which may be comprehensive enough to mimic (mostly) aspects of human intelligence, particularly in the arena for which programmed. A 2AAA may indeed be the "smartest" crew member, by an order of magnitude or two, but the "dumbest" crew member will do a better job writing a novel. A 2AAA could mimic emotion, and mimic appropriate responses to emotional displays, but it has none, nor could it really understand existential angst.

I have taken here the devil's advocate in arguing for LBB8's system, giving a perspective under which it all makes sense. Numbers-wise, does it really screw up LBB1-5? Oh, yes. Do they themselves (LBB1-5) always make sense? Different conversation. :nonono:

Here's the rub: LBB8 is, given the current expotential trajectories of robotics and computational capability, very conservative. The Accords give us a nice fig leaf for our minimizing the role of robots. LBB8 gives us a glimpse into their potential.

By way of another analogy, the selective breeding of humans could, scientifically, have huge potential. Because of historical experience and our own moral system, however, we find the mere thought repugnant. This does not mean that it would not be effective in some dimensions. But, "What does it profit a man..." All that is technologically possible is not morally palatable: indeed, as our technological might increases, the proportion of that of the possible that is acceptable plummets.

 

[samuelvss]

With the rules from book 8 You can also teach the robots skills. Quite high level of skills... And this means that You can have robots as expert gunners or fighter pilots... ***{see below}

But this would of course mess up the strict starship automation rules...

Well, maybe. When does the 3I allow robot gunners?

It does not mess up the rules if it is not present. The issue may be that humans are cheaper, in some cases. That being said, trying to hire 12 gunner-4's would be a bitch; buying them is technologically possible. What are the law levels on "Gunner" software, though? It may, for reasons readily apparent, be illegal in the 3I. If the IN does not use robot fighters, or robot-assisted fighters, for cultural reasons, but there are huge tacitical advantages, then outlawing them completely might be a logical (or the only
logical) response.


***You mean the 1C Gunner4 !!
Astromech droid. Gunner and basic rescue functions.
URP: 82203-04-LQ226-PF4L(L)
TL 13, kCr 243, 278 kg 25yr note @ kCr 1.54/mo.
120 Liters Lhyd fuel gives 50 hr endurance (closed) or
500 hours endurance (open). Work stations typically have
power run to them.
All systems will typically not run simultaniously, so endurance
will typically be much greater. E.g.: Cannot walk use arms and all tools
at once.
Gunnery-4, Security-2, Mech/Elec/Med/Rescue/Commo/Ship's Boat-1

[NOTE: This is the bargain Gunner4!! :smirk:]

Note: For all imperial starships with a droid that fills any crew station, the droid is required to have a minimum of 1 level of Commo, Medical & Rescue, plus one level of the appriate skill for the emergency craft (ship's boat or gravitics)

--------------------------------------------------------------------------

 

[atpollard]

With the rules from book 8 You can also teach the robots skills. Quite high level of skills... And this means that You can have robots as expert gunners or fighter pilots...

Note that the key here is "apparent intelligence." A 2AAA, with a superhuman "apparent intelligence," is still only semi-autonomous. There are things, by way of human nature, that can not be programmed.

By way of analogy, my [crappy little] computer is a better speller and mathmatician than I am, by orders of magnitude. I could program a computer to ace an IQ test. I can create an "expert system," that reacts tactically. By some measures of IQ then, my computer with the right software is already smarter than me. It, of course, could not write this email. As TL's increase and more automation is sought in the drive towards AI, more areas in which intelleigence can be measured will be filled up.

On a 'Thought Experiment' level:

I would question the meaning of 'Skill Level', 'Semi-Autonomous', and 'Apparent Intelligence' as they are applied to this problem. Let's use a 'Robot Gunner' as an example. It seems reasonable that a computer brain could be created that could react faster and hit the target a higher percentage of the time than a Human Gunner - like the Phalanx Ship defense system. But what else does being a 'Gunner' entail. Since 1 or 2 skills are gained per 4 years of character generation (and improved only once in an 8 year period after character generation) it would seem that the skill Gunner-1 represents the total of all experience and theory gained in 2 to 4 years of performing a task as a vocation.

So what might a Gunner do in 2-4 years of work?

• Practice hitting a target.
• Maintain his weapon
• Calibrate the interface
• Identify malfunctions, and repair them.
• Gain practical experience in complex situations involving Identifying targets.

While a robot could certainly excel at 'hitting a target', could it accurately identify a possible target as friend or foe? Will it constantly shoot down satelites that are on a potential intercept course? Other armed ships at the starport? Can it identify that the fighter next to the communication sattelite is out of place and a potential threat? Would it attack a launch servicing the sattelite? Could it identify a possible ECM Ghost or will it shoot all your missiles at every blip the enemy sends your way?

This is where terms like 'semi-autonomous' and 'high apparent intelligence' fail to properly apply to the concept of a skilled professional Gunner. I can accept that 'high apparent intelligence' means that the robot can learn to shoot the wings off a fly from orbit, but 'semi-autonomous' rather than 'full AI' suggests that the SPA security guards at the starport are in great danger - just like the ship attempting to land on the next pad.

 

[samuelvss]

On a 'Thought Experiment' level:

I would question the meaning of 'Skill Level', 'Semi-Autonomous', and 'Apparent Intelligence' as they are applied to this problem.

. . . . .

So what might a Gunner do in 2-4 years of work?

• Practice hitting a target.
• Maintain his weapon
• Calibrate the interface
• Identify malfunctions, and repair them.
• Gain practical experience in complex situations involving Identifying targets.

While a robot could certainly excel at 'hitting a target', could it accurately identify a possible target as friend or foe?

. . . . . .

I can accept that 'high apparent intelligence' means that the robot can learn to shoot the wings off a fly from orbit, but 'semi-autonomous' rather than 'full AI' suggests that the SPA security guards at the starport are in great danger - just like the ship attempting to land on the next pad.

For the most part, you're preachin' to the choir.

First, semi-autonomous means to me, "Not autonomous." That being said, I think your thought experiment mischaracterizes some items of skill and autonomy; speaking, therefore, to the 5% of what I disagree with rather than the 95% that I accept, I set out my objections:

By way of analogy, a U.S. tank crew: I talk about the tanks up to the M1A1, as the A2 is different. The Tank Commander (TC), exercises, or must be able to exercise, an autonomous role. He directs movement (the driver drives), he designates targets (the gunner shoots), and the ammunition to be used (the loader loads), as well as interacting with the other elements outside the tank, human or otherwise, that require his attention, such as his next higher leader, wingman, local "crunchies" [the CDAT's term for we poor dismounts that, among their other social failings, inspire us to dub them Computerized Dumb-Ass Tankers], aviation, fire support, and other scores of other potential entities, and variables going into a couple of orders of magnitude. A semi-autonomous robot could not do this job.

The gunner, however, takes the given target, and engages. He doesn't just decide to go shoot something, or even which something to shoot without a specific order to do so. He does not shoot things inside the motorpool; not simply because he is a sentient being with hopes and dreams, but because, just like the "R2D2's" on the USS Nimitz, his targetting is turned off; indeed, there are a few failsafes. The loader is further responsible that the weapons are unloaded accept where they are ordered to be otherwise. These are roles suitable for the semi-autonomous. They are not deciding on when to shoot, or when to go where, but rather using their skills to follow the orders they are given.

In some activities, human "intelligence" is over-rated. Most aviation accidents are not a lack of human intelligence, as a problem-solving variety, but rather either a conscious decision to ignore protocols, or forgetting to follow such protocols.

So while I agree that there are significant limitations that robots would have in the OTU, I think that their strengths can balance those in the proper circumstances. Which circumstances require human guidance. If a captain or chief gunner tells the 2A gunner to shoot whenever its seems appropriate, o: then they themselves are falling significantly short of the AI threshold.

We surround ourselves with machines to compensate for our weaknesses. They certainly have weaknesses of their own. If we survive, it is because we get the balance right. That is autonomous.

 

[mbrinkhues]

From reading B8 I get the feeling that "semi autonomous" there is more "not having / not having much of a creative thought process". The systems can react to situations within their programming and even interpolate when encountering new situations. Without that bots like the Rashush personal servants (IIRC TL12) or similar units would not work.

With that you get a very solid PC replacement in these brains, even more so in combat tasks up to a 1Lt / Captain (Who are generally doing things the "Approved and official military way" read: Acting semi-autonomous)