Announcements

See our latest update! Next Steps for Building City of Titans!

Get the latest Patch now! The MacOS Client Launcher is Now available in the store! If you've already donated $50 or more, you'll find the Mac Launcher on your rewards page. https://store.missingworldsmedia.com/

Join the conversation on Discord: https://discord.gg/w6Tpkp2

Customizing AI

30 posts / 0 new
Last post
Halae
Halae's picture
Offline
Last seen: 9 months 4 days ago
11th Anniversary Badge
Joined: 09/17/2014 - 09:37
Customizing AI

I play a lot of Guild Wars 2, and I've been excited for the upcoming expansion to that game; well, it's an expansion. Who wouldn't be excited? A thing that caught my eye, however, was the AI routine changes they're pushing into the game. You can watch the video for that Here if you like.

The reason I bring this up, however, is due to my love of Master type characters in games, but so very few games actually give you the ability to manage what your minions are doing. Back in CoX, whenever we had minions we could either let them be aggressive, defensive, or passive, and while that wasn't bad, per se, having only three commands could be stifling. People talk about creativity for their minions. New outfits, different appearances made using the character creator, that sort of thing. What I'm talking about, though, is AI customization.

The point behind the idea is simple, and has actually been done before by such games as Final Fantasy 12 and Dominions 1 through 4 (which I sadly don't have a link to show off). Basically, you can classify where your AI companions stand, how far, when they should use their various abilities, and so on. All it is is basically editing AI routines using a dropdown menu that accounts for the majority of variables. This could add a tactical level of gameplay to it, and for those that like to do so, they could set things up such that they don't have to lift a finger, just observe the well oiled machine that is their minions moving forward through a fight. The real challenge, for them, is changing battle strategies based on types of enemies.

Noteworthy is that this wouldn't be required. Default AI routines would be similar to the ones used by CoX for their Mastermind system, which was a little basic but hey, it worked. This would give an additional layer of depth for those that want it without excluding the impatient folks who just want to watch robots smash things.

One thing to keep in mind is that people will break this sort of thing, and not in the way where they become overpowered. I'm talking making it so it doesn't work anymore - as such, a "Reset To Default" button is recommended. Another good choice would be the ability to save your build into a template folder you could also load from, sort of the same way you could save and load costumes in the CoX character creator. The reasoning for this is because you'd want different AI setups for, say, a Giant Monster than you would an Archvillain or a group of trash mobs. This would encourage sharing of ideas with their builds among the community, and still let those who don't want to put too much thought into things download an available build and slot it in, since the AI routine they want has already been developed by another player.

In addition to those QoL ideas, I can think of a couple others. The first is saving multiple routines - perhaps three slots, which default to passive, defensive, and aggressive. This means people wouldn't have to constantly replace their AI routines every time they want to change them, just swap over to one of three saved ones, which could even be done in combat, if necessary.

The second idea would be a "command" setup. As in, you can attach macros to certain abilities, setting up your minions to do certain things. for example, perhaps you have a "Stun!" button, where you command your minions to use a stunning ability on your target. This'd give you an additional level of on-demand control that the AI routines don't account for, which could be important.

Anyways, these are mostly just ideas based on things other games are doing. I'd be happy with the inclusion of any viable and fun minion system, but these ideas would give an additional level of depth to what you can do with your minions, and give those people who enjoy setting things up and then watching them go like a Rube Goldberg machine (like myself) something awesome to do.

An infinite number of tries doesn't mean that any one of those tries will succeed. I could flip an infinite number of pennies an infinite number of times and, barring genuine randomness, they will never come up "Waffles".

Izzy
Izzy's picture
Offline
Last seen: 3 years 5 months ago
kickstarter11th Anniversary Badge
Joined: 10/09/2013 - 11:09
++++++11111111

++++++11111111

"I bet my Pet AI will be better than yours."
As I've stated a few times before concerning this subject. >:-]

Tannim222
Tannim222's picture
Offline
Last seen: 5 hours 6 min ago
Developer11th Anniversary Badge
Joined: 01/16/2013 - 12:47
Customizing AI while

Customizing AI while interesting would be a long, long way off if we were ever to implement it. One of the reasons why the Operator Classification was put so high on the list for our KS was due to the system's needed to create the kind of AI our AI Lead hopes to make for the game. One where they AI isn't run via scripted events but learned behaviors. As such the commands players have access to would inform the pet AI what they player desires and the AI would then go about performing that action in a way that it self deems best fit for that situation (if there is more than one way to skin a cat in a given situation).

Giving players control over when and how often what abilities are triggered would place a heavy burned on devs for ensuring consistent bounds of performance. Which is why I said it would be a long way off if ever. I do hope to impart something unique which is a step away from the MM AT but it remains to be seen if it is possible and how it would impact design. Along the lines of your Stun button suggestion it would be more likened to additional commands to inform the pet of the desired style of play. Where as before there were three basic commands, in addition there would be a Something like a Support command, control command, which would let the AI give preference to any abilities which are consistent with those commands. Take it as something of a mix between the on demand style and the behavioral routine style of player control. These additional command controls however are dependent upon how versatile we will be able to make the various pets themselves.

Keep in mind the MM AT one already one of the most versatile high performing ATs of the old game. Giving players in depth control could actually be worse for the health of the class not better in that the top end of performance possibilities would have to be taken into account which means those playing with high skill would see many benefits of the class, but those who didn't do so might feel the class under perform and is basically skewed in difficiculty (or general players end up feeling 'forced' to play the 'hard way' to play well). Not to mention how much players could actually achieve beyond the basic commands with creating their own macros. I was one of those who could on the fly, control each pet individually, in groups by rank, in groupings by pet name, and while I couldn't tell a pet what ability to fire off when (and there are a ton of valid reasons for this to not be allowed), the deeper levels of control via macros made functionality operate like night a day from standard controls.

Now we do hope to help ease the way players can access the wider array of possible macros, key binds, and commands for any and every Classification with a nifty UI menu system which should help further with players entering the more in depth operations of the Operator.


I don't use a nerf bat, I have a magic crowbar!
- Combat Mechanic -
Tech Team.
Halae
Halae's picture
Offline
Last seen: 9 months 4 days ago
11th Anniversary Badge
Joined: 09/17/2014 - 09:37
I figured i might get a

I figured i might get a response like that; as I recall, Operators aren't even scheduled to be in the launch of the game, right? Because they're extremely work intensive compared to other archetypes.

When I make suggestions on this forum they're less "I want this done" and more "I want this to be considered" because while most of my ideas are pipe dreams, there's always things to take from ideas, even nonviable ones. I definitely like the idea of having the ability to choose between combat styles on the fly though - perhaps not as robust as an AI customization, but it gives you a broader sense of being able to handle the minions.

But everything depends on everything else, right? Thanks for responding; it's always lovely to see a dev respond regarding what is and is not viable from a development standpoint.

And that extra UI thingamadoodle sounds awesome.

An infinite number of tries doesn't mean that any one of those tries will succeed. I could flip an infinite number of pennies an infinite number of times and, barring genuine randomness, they will never come up "Waffles".

Greyhawk
Greyhawk's picture
Offline
Last seen: 1 year 5 months ago
11th Anniversary Badge
Joined: 01/03/2015 - 19:17
Tannim222 wrote:
Tannim222 wrote:

Customizing AI while interesting would be a long, long way off if we were ever to implement it. One of the reasons why the Operator Classification was put so high on the list for our KS was due to the system's needed to create the kind of AI our AI Lead hopes to make for the game. One where they AI isn't run via scripted events but learned behaviors. As such the commands players have access to would inform the pet AI what they player desires and the AI would then go about performing that action in a way that it self deems best fit for that situation (if there is more than one way to skin a cat in a given situation).

...wow.

If my Mastermind pets had been given the ability to learn based on past inputs my Mastermind character would have been completely impossible to defeat within a few months of achieving level 50. Even simple learned behaviors can dramatically change outcomes in conflict scenarios. Just watch chimpanzee troops encountering predators sometime.

An Operator with pets that learn. Very scary idea. Groundbreaking, even. I don't think the gaming world is ready for this yet.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
My author page at Amazon: https://amzn.to/2MPvkRX
My novelty shirts: https://amzn.to/31Sld32

TTheDDoctor
TTheDDoctor's picture
Offline
Last seen: 1 month 1 week ago
11th Anniversary Badge
Joined: 03/29/2014 - 15:26
Greyhawk wrote:
Greyhawk wrote:

Tannim222 wrote:
Customizing AI while interesting would be a long, long way off if we were ever to implement it. One of the reasons why the Operator Classification was put so high on the list for our KS was due to the system's needed to create the kind of AI our AI Lead hopes to make for the game. One where they AI isn't run via scripted events but learned behaviors. As such the commands players have access to would inform the pet AI what they player desires and the AI would then go about performing that action in a way that it self deems best fit for that situation (if there is more than one way to skin a cat in a given situation).

...wow.
If my Mastermind pets had been given the ability to learn based on past inputs my Mastermind character would have been completely impossible to defeat within a few months of achieving level 50. Even simple learned behaviors can dramatically change outcomes in conflict scenarios. Just watch chimpanzee troops encountering predators sometime.
An Operator with pets that learn. Very scary idea. Groundbreaking, even. I don't think the gaming world is ready for this yet.

I think the reason they didn't release the masteries for the Operator is because they needed to level the playing field for the other classifications. XD

<==========)===O|TtDd|O===(==========>
My original character profiles!

Redlynne
Redlynne's picture
Offline
Last seen: 10 months 1 week ago
kickstarter11th Anniversary Badge
Joined: 10/28/2013 - 21:15
I'm just worried how much of

I'm just worried how much of a keybind and UI nightmare this sort of thing would become simply in order to be able to CONTROL what your Pets are likely to be doing.

Let's start off with the keybind side of things first.

In City of Heroes, there was the Mastermind Keybind Files "tree" of keybinds that could be used to control your Pets. In my case, I think it was something like *10* different keybind files, all driven by use of the NUMPAD 0-9, including use of = / * - + and the decimal point too for addressing individual Pets and dismissing them. It got VERY complex, VERY quickly! It was the work of an afternoon to write it all up and then load them up and test proof them in-game to make sure there weren't any typos or errors in the commands. It was an INVESTMENT to be able to do this, and although there were Guides written for how to do it, it was still a very finicky process to actually go ahead and actually DO it and PROOF it.

Kheldians who didn't limit themselves to Human Form encountered a similar issue, but almost an order of magnitude smaller and easier to manage. Tri-form Kheldians also effectively required a "tree" of 4 Keybind Files for controlling their UI in their various forms.

So when you talk about adding Stances to our ability to control Pets, I just look at my NUMPAD and ask ... where am I going to put all that? I mean, my Mastermind Keybind Files were already using *16* keys on the NUMPAD just for the address and command functions, and you want to add on EVEN MORE? Where am I going to have room for all of that?

So let's just put down a marker right now.
If you need more than 16 buttons to do all of the COMBINATIONS of control commands for Pets, you're doing it wrong.

Pets to address selection:
Numpad 0 : All Pets
Numpad 1 : All Minions
Numpad 2 : All Lieutenants
Numpad 3 : All Bosses
Numpad + : Lieutenant 1
Numpad - : Lieutenant 2
Numpad = : Minion 1
Numpad / : Minion 2
Numpad * : Minion 3

Commands to Pets selection:
Numpad . : Dismiss (I usually included a Shift with this command for "accident prevention")
Numpad 4 : Aggressive
Numpad 5 : Passive
Numpad 6 : Defensive
Numpad 7 : Attack Target
Numpad 8 : Go To
Numpad 9 : Follow

And that's not even including anything involving getting your Pets to do any emotes or chatting. For me, those two things were a whole EXTRA LAYER of duplication written into the keybind files to allow that kind of functionality to happen quickly (think Shift and Ctrl key combinations to access the extra layers of functionality in combination with the Numpad keys).

So to put it simply ... there just isn't any more "room at the inn" when it comes to being able to issue more complex sets of commands to Pets. The UI at the (wo)man/machine interface gateway is ... FULL!

Now, that said ... there is a design space that we might be able to leverage for this sort of customization, but it's going to have to happen somewhere else than the keybinds. Just about the only thing I can think of along these lines would be to give us the capacity to "weight" the importance of allowing Pets to commit certain actions.

As any Robots/* Mastermind could tell you, in City of Heroes, every Pet (not just Robots) had the ability to Brawl ... even if Brawling was a suicidally STUPID thing for them to do. It meant that a Pet such as Robots, which had ALL Ranged Attacks would occasionally just run up to melee an opponent FOR NO GOOD REASON and then stay parked in melee FOR NO GOOD REASON rather than returning to a safe distance from the fray. This behavior was especially egregious against Hamidon Mitochondria where staying at a distance was literally the difference between Life and Debt. Paragon Studios tried a variety of different "fixes" over the years, and pretty much none of them worked, because of how the AI was programmed. It was just too ... rigid ... in its decision processes for them to ever effectively discard Brawl for those Pets that outgrew the need to Brawl at all.

So ... what I'm wondering is if there could be a UI window in the game which allows the Player to set the imperative/priority levels for all of the attack Powers that Pets will have, so as to be able to "manage" (in advance and out of combat) what sorts of behaviors Pets ought to have when they are engaged in combat. Let's look at Robots as an example for what I've got in mind:

Battle Drones:
Basic: Smash (melee, single target), Laser Blast (ranged, single target)
Upgrade 1: Heavy Laser Burst (ranged, single target)
Upgrade 2: Full Auto Laser (cone, volume of effect)

Protector Bots:
Basic: Laser Blast (ranged, single target), Force Shield (ranged, single target)
Upgrade 1: Heavy Laser Burst (ranged, single target), Repair (ranged, single target)
Upgrade 2: Photon Grenade (ranged, target volume of effect), Seeker Drones (spawn pet)

Assault Bot:
Basic: Smash (melee, single target), Plasma Blast (ranged, single target)
Upgrade 1: Dual Plasma Blast (ranged, volume of effect), Flamethrower (cone, volume of effect)
Upgrade 2: Swarm Missiles (ranged, volume of effect), Incendiary Missiles (ranged, volume of effect)

Looking at this list, you can see that there are essentially 4-6 Powers available to each Pet (when fully upgraded), and that each Pet starts off with 2 Powers. Here's what I'd want to do to "control" how those Powers get used:

Positional Priority
0 = IF not within range to use Power THEN do not move into range to use Power
1 = IF not within range to use Power THEN move into range to use Power IF no other Power is available to be used at current range to Target
2 = IF not within range to use Power THEN move into range to use Power AND use Power once within range

Usage Priority
0 = Do not use Power
1 = Use Power if a higher priority Power is not available for use
2 = Use Power whenever possible

This would create a quick & dirty "matrix" for how Powers are meant to be employed by specific Pets. So let's then take a look at what it would mean in a context like a Robotics/* set. I personally would program something like this through the UI window:

Battle Drones:
Smash: P1, U1 (initially, before upgrades) ... P0, U1 (later on after achieving upgrade levels)
Laser Blast: P2, U2
Heavy Laser Burst: P2, U2
Full Auto Laser: P2, U2

Protector Bots:
Laser Blast: P1, U1
Force Shield: P2, U2
Heavy Laser Burst: P1, U1
Heal: P2, U2
Photon Grenade: P1, U1
Seeker Drones: P1, U1

Assault Bot:
Smash: P0, U1
Plasma Blast: P1, U1
Dual Plasma Blast: P2, U2
Flamethrower: P1, U1
Swarm Missiles: P2, U2
Incendiary Missiles: P2, U2

So what does all of that "say" about how I, as a Player, would play a Robotics/* Mastermind? Well, it would mean that "Brawling" would be an extremely low priority for the Pets that could do it. If they're already in range to do it, go ahead, but don't deliberately move into range to do it. That then gives me, as the Player, the OPTION to command my Pets to Go To a location where they can melee (or conversely, the Foes close into melee range with my Pets on their own) ... in which case my Pets would use their Smash Power, but that's the only conditions under which it would get used. Everything else falls into either a "use it if you can" or a "get into position to use if you can't do anything else just then" sort of set of imperatives. You'll note that for the Protector Bots I prioritized their defensive over their offensive Powers, for example.

I figure that this sort of "decision bias/weighting" of the AI for Pets is something that it would be wise to have built into the game from a pretty early state of development, so it can be used not just for Operator Pets but also for the Foe NPCs in general to help fine tune their behaviors. If nothing else, this sort of thing ought to help counter the propensity of EVERY Foe NPC in every game to simply bumrush the PCs in every situation.

You could even add a "Pursuit" parameter to things on top of the above, referencing Line of Sight.

Pursuit Priority
0 = IF Line of Sight is blocked THEN do not move
1 = IF Line of Sight is blocked THEN move to close range IF Target is within range of a recharged attack Power (ignoring Line of Sight for range)
2 = IF Line of Sight is blocked THEN move to close range with Target

This would then create a set of behaviors that would cause (or prevent) Pets from running off in pursuit of Foes. It would create conditions in which Pets (and Foe NPCs) would be more resistant to abandoning their positions, meaning that lures and "pulling" tactics may or may not be effective against them.

If you want to allow for more complex permutations of behavior, you can increase the range of options from 3 (0, 1, 2) to 5 (0, 1, 2, 3, 4) so as to allow a greater degree of fine tuning for behavioral models. Note that this sort of thing would be relatively easy to program as an in-game UI window element using sliders for each of the Powers that a Pet can use. The combination of settings would then be determinative for the "types" of behaviors the Pet would typically exhibit while in combat and guide its AI in its decision processes.

To quote a few chess masters ... YOUR MOVE ...


Verbogeny is one of many pleasurettes afforded a creatific thinkerizer.
Greyhawk
Greyhawk's picture
Offline
Last seen: 1 year 5 months ago
11th Anniversary Badge
Joined: 01/03/2015 - 19:17
Hmm...how to better customize

Hmm...how to better customize pet commands? Tough call.

Myself, I always wanted to not waste slots on my power bar on pet summons and dismissal. Most of the time, my summon commands were on a separate bar accessed by "Shift+PowerBarNumber". Usually Shift+4, but above level 30 or so, Shift+5, especially after the barrage of temporary powers received through Veteran Rewards. Near the end those filled up Shift+4 all by themselves. So if "Summon Pet Such&Such" could exist as a simple keystroke rather than a normal power slot (or something), it would help me greatly.

Pets for Operators will need at least two tiers of commands: "all pets" and "tier# pets". Basic powers (the same power commands other characters use) rather than being used for summons could perhaps be used for "all pets" commands. Assuming a standard character has 10 powers:

1: All Pets Attack
2: All Pets Retreat, Passive
3: All Pets Retreat, Defensive
4: All Pets Attack from melee range
5: All Pets Attack from distance
6: Operator uses personal attack A
7: Operator uses personal attack B
8: Operator uses support power A (pets gain improved stats)
9: Operator uses support power B (pets gain improved stats + a special power)
0: Operator uses Thematic power (determined by pet types, sometimes a "nuke", sometimes a huge "heal all", sometimes a short lived pet swarm)

F1: Tier 1 Pets Attack
F2: Tier 2 Pets Attack
F3: Tier 3 Pets Attack
F4: Tier 1 Pets Attack from melee range
F5: Tier 2 Pets Attack from melee range
F6: Tier 3 Pets Attack from melee range
F7: Tier 1 Pets Attack from distance
F8: Tier 2 Pets Attack from distance
F9: Tier 3 Pets Attack from distance
F10: Dismiss and Resummon all pets

Numpad_1: Tier 1 pets Retreat, Defensive
Numpad_2: Tier 2 pets Retreat, Defensive
Numpad_3: Tier 3 pets Retreat, Defensive
Numpad_4: Tier 1 pets use special power
Numpad_5: Tier 2 pets use special power
Numpad_6: Tier 3 pets use special power
Numpad_7 through Numpad_9: reserved for player binds/macros

Special commands accessed by creating binds/macros/typed commands:

AllPetsDefendLocation(x, y) (x, y as distance from Operator)
Tier1PetsDefendLocation(x, y)
Tier2PetsDefendLocation(x, y)
Tier3PetsDefendLocation(x, y)

AllPetsAttackLeftFlank (left and right from Operator's perspective)
Tier1PetsAttackLeftFlank
Tier2PetsAttackLeftFlank
Tier3PetsAttackLeftFlank

AllPetsAttackRightFlank
Tier1PetsAttackRightFlank
Tier2PetsAttackRightFlank
Tier3PetsAttackRightFlank

AllPetsSearchMap (for enemies, random direction from current location, used only on mission maps, first encountered enemy location appears on Operator's mini-map, pets do not attack, pets return to Operator after locating first enemy/enemy group)
Tier1PetsSearchMap
Tier2PetsSearchMap
Tier3PetsSearchMap

AllPetsSearchLeft (for enemies, used only on mission maps, first encountered enemy location appears on Operator's mini-map, pets do not attack, pets return to Operator after locating first enemy/enemy group)
Tier1PetsSearchLeft
Tier2PetsSearchLeft
Tier3PetsSearchLeft

AllPetsSearchRight (for enemies, used only on mission maps, first encountered enemy location appears on Operator's mini-map, pets do not attack, pets return to Operator after locating first enemy/enemy group)
Tier1PetsSearchRight
Tier2PetsSearchRight
Tier3PetsSearchRight

AllPetsSearchAndAttack (for enemies, random direction from current location, used only on mission maps, first encountered enemy location appears on Operator's mini-map, pets attack first encountered enemy/enemy group)
Tier1PetsSearchAndAttack
Tier2PetsSearchAndAttack
Tier3PetsSearchAndAttack

AllPetsSearchAndAttackLeft (for enemies, used only on mission maps, first encountered enemy location appears on Operator's mini-map, pets attack first encountered enemy/enemy group)
Tier1PetsSearchAndAttackLeft
Tier2PetsSearchAndAttackLeft
Tier3PetsSearchAndAttackLeft

AllPetsSearchAndAttackRight (for enemies, used only on mission maps, first encountered enemy location appears on Operator's mini-map, pets attack first encountered enemy/enemy group)
Tier1PetsSearchAndAttackRight
Tier2PetsSearchAndAttackRight
Tier3PetsSearchAndAttackRight

Just for fun:
AllPetsUse (emote name)
Tier1PetsUse (emote name)
Tier2PetsUse (emote name)
Tier3PetsUse (emote name)

Although this seems simple and straightforward to me, I realize many people in today's world don't really need discrete commands like this. Many people would be very comfortable with a simple scripting language that broke each of these discrete commands down into components and then the Operator could string the components together to achieve the desired result.

AllPets()
TierPets()
Use()
Defend()
Attack()
Search()
FlankAttack()
VectorAttack()

Then the pieces could be cobbled together something like this:

TierPets(1,Defend(3,2))
TierPets(2,Use(cheer))
TierPets(3,FlankAttack(r))

To prevent someone from entering a map, standing by the doorway, and dispatching their pets to clear the map, simply do not let the pets advance beyond the first enemy they find after receiving a "search" or "search and attack" command. Also, if the survivability of the pets is greatly diminished by being beyond a certain distance from the Operator, then the Operator would be forced to move to the pet location as soon as contact is made with an enemy. The idea is to provide enough choices to allow for more creative tactical thinking then simply, "Attack" or "Defend", while not actually requiring those choices to be made. If a player wishes, they can simply select an enemy, hit "1", then move on to the next enemy and do the same. Or, if they prefer, they can compose elaborate search and destroy keybinds, send their pets to the flank while their team attacks head on, use a pet from any tier to pull back to a location the other pets are defending, or use their top tier pet to tank while directing the other pets to attack from the flanks. In a pinch they can direct each tier to use it's special power (especially useful when only one tier has a heal or buff power that it never seems to bother using) regardless of how each type of pet defines that power.

In this way Operators will be able to actually "operate" their pets.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
My author page at Amazon: https://amzn.to/2MPvkRX
My novelty shirts: https://amzn.to/31Sld32

Redlynne
Redlynne's picture
Offline
Last seen: 10 months 1 week ago
kickstarter11th Anniversary Badge
Joined: 10/28/2013 - 21:15
The more ... layers ... of

The more ... layers ... of command options you stack up, the more complex and time consuming it becomes to successfully execute those commands in a timely fashion. One And Done commands typically tend to work adequately on their own for their intended purpose. Two Step Combos can require quick thinking and decent timing. Three Step Waltzes are almost always too much fancy footwork to pull off reliably in real time. Things only get worse from there.

At most, you want to be working with only 2 steps ... WHO and DO ... which in the majority of cases ought to effectively wind up skipping over the "WHO" step, since there's no change there since the last command was issued and it just carries over, and you simply execute the DO portion of it.

That's why almost all of the Mastermind Keybind schemes that I saw involved 9 keybinds for the WHO portion, but only 7 keybinds for the DO portion (including Dismissal as the 7th). Think about that. 9 keybinds to Target Select any of a combination of 6 Pets.

First you choose WHO ... and then you command them what to DO. I'm hard pressed to think of a better organizational scheme at the keybind level than that.


Verbogeny is one of many pleasurettes afforded a creatific thinkerizer.
Gangrel
Offline
Last seen: 3 years 6 months ago
11th Anniversary Badge
Joined: 09/15/2013 - 15:14
Greyhawk wrote:
Greyhawk wrote:

*snip*

I read all of that and I went: And people say that Eve Online has a steep learning curve, and that Elite Dangerous uses too many key presses (100ish).

That just blew that all out of the water......

Quote:

1) I reject your reality.... and substitute my own
2) Not to be used when upset... will void warranty
3) Stoke me a clipper i will be back for dinner
4) I have seen more intelligence from an NPC AI in TR beta, than from most MMO players.

Halae
Halae's picture
Offline
Last seen: 9 months 4 days ago
11th Anniversary Badge
Joined: 09/17/2014 - 09:37
Any time you're given

Any time you're given potential control over something, people will take it to the furthest extent possible. I was one of those schmucks that never bound key presses like this, preferring to just actually play the game rather than fiddle with it until I had everything down pat. Never really had any problems because of that, but I could see how I could have improved by doing more to control my demons.

An infinite number of tries doesn't mean that any one of those tries will succeed. I could flip an infinite number of pennies an infinite number of times and, barring genuine randomness, they will never come up "Waffles".

Gangrel
Offline
Last seen: 3 years 6 months ago
11th Anniversary Badge
Joined: 09/15/2013 - 15:14
Halae wrote:
Halae wrote:

Any time you're given potential control over something, people will take it to the furthest extent possible. I was one of those schmucks that never bound key presses like this, preferring to just actually play the game rather than fiddle with it until I had everything down pat. Never really had any problems because of that, but I could see how I could have improved by doing more to control my demons.

Of course, I can also right now see how I can utilise Voice Attack to do stuff like this all without actually having to press a key and give the commands just with my voice.

Quote:

1) I reject your reality.... and substitute my own
2) Not to be used when upset... will void warranty
3) Stoke me a clipper i will be back for dinner
4) I have seen more intelligence from an NPC AI in TR beta, than from most MMO players.

Redlynne
Redlynne's picture
Offline
Last seen: 10 months 1 week ago
kickstarter11th Anniversary Badge
Joined: 10/28/2013 - 21:15
Good for you, Gangrel.

Good for you, Gangrel. Unfortunately, some players have hearing problems, up to and including deafness ... so, I regret to say, use of Voice Attack can't be a one size fits all solution to the problem. It may work fine for some, but there will be a subsection of Players for whom it will not work.

I'm not nitpicking with this so much as covering all the bases. Don't want to fall into the trap of thinking that just because *I* have a sandwich, no one else could ever go hungry.


Verbogeny is one of many pleasurettes afforded a creatific thinkerizer.
Halae
Halae's picture
Offline
Last seen: 9 months 4 days ago
11th Anniversary Badge
Joined: 09/17/2014 - 09:37
If I recall, that program

If I recall, that program also requires you to enunciate clearly, meaning you thick accents would pose problems as well.

An infinite number of tries doesn't mean that any one of those tries will succeed. I could flip an infinite number of pennies an infinite number of times and, barring genuine randomness, they will never come up "Waffles".

Gangrel
Offline
Last seen: 3 years 6 months ago
11th Anniversary Badge
Joined: 09/15/2013 - 15:14
Halae wrote:
Halae wrote:

If I recall, that program also requires you to enunciate clearly, meaning you thick accents would pose problems as well.

It uses the windows speech engine, so it actually learns over time and improves its recognition. The most common reason as to why it fails is that people just pick it up and expect it to work out of the box, without training.

In that case, it will fail.

But the best part is that you can also tell Voice Attack *what* to listen for and what to do when it hears the phrase, so even if it *consistently* mishears a word, you can just add in what it "hears" instead into the command so that next time it does the correct action.

It won't actually do anything without having a command structure initially set up.

To be honest, I only made the comment in a case of of what I actually said, and not really in context of what I had quoted. That was my mistake on the part.

Hell, it can even work without a voice commands.... although at that point in time it then becomes a macro system (although one that you can also do some very basic logic tracking in on top)

Quote:

1) I reject your reality.... and substitute my own
2) Not to be used when upset... will void warranty
3) Stoke me a clipper i will be back for dinner
4) I have seen more intelligence from an NPC AI in TR beta, than from most MMO players.

TTheDDoctor
TTheDDoctor's picture
Offline
Last seen: 1 month 1 week ago
11th Anniversary Badge
Joined: 03/29/2014 - 15:26
Halae wrote:
Halae wrote:

but I could see how I could have improved by doing more to control my demons.

The fact that your avatar has horns and red glowing eyes made my sides explode when I read this. ^

<==========)===O|TtDd|O===(==========>
My original character profiles!

Redlynne
Redlynne's picture
Offline
Last seen: 10 months 1 week ago
kickstarter11th Anniversary Badge
Joined: 10/28/2013 - 21:15
Dragging this back towards

Dragging this back towards something resembling Players manipulating the AI more directly/completely/discretely, having thought about the matter further, I'm still convinced that we both don't, and shouldn't, upset the applecart of attack/go to/follow and aggressive/passive/defensive. Keeping THAT in place ought to be paramount!

The only thing I wish we'd been able to "fine tune" better in City of Heroes was the ... priorities ... our Pets had with respect to using their Powers. Being able to prioritize (as I posted previously) as a sort of "programming" of Pets chosen in advance by the Player would go pretty much the rest of the way towards enabling Players to exercise greater control over how their Pets "behave" in combat, in ways that could potentially open new strategies, tactics and playstyles beyond the standard/obvious.


Verbogeny is one of many pleasurettes afforded a creatific thinkerizer.
Gangrel
Offline
Last seen: 3 years 6 months ago
11th Anniversary Badge
Joined: 09/15/2013 - 15:14
Redlynne wrote:
Redlynne wrote:

The only thing I wish we'd been able to "fine tune" better in City of Heroes was the ... priorities ... our Pets had with respect to using their Powers. Being able to prioritize (as I posted previously) as a sort of "programming" of Pets chosen in advance by the Player would go pretty much the rest of the way towards enabling Players to exercise greater control over how their Pets "behave" in combat, in ways that could potentially open new strategies, tactics and playstyles beyond the standard/obvious.

Very handy if this prevented the AI from going to melee range to use a ranged attack....

Quote:

1) I reject your reality.... and substitute my own
2) Not to be used when upset... will void warranty
3) Stoke me a clipper i will be back for dinner
4) I have seen more intelligence from an NPC AI in TR beta, than from most MMO players.

Tannim222
Tannim222's picture
Offline
Last seen: 5 hours 6 min ago
Developer11th Anniversary Badge
Joined: 01/16/2013 - 12:47
That won't be an issue if we

That won't be an issue if we using heuristic driven AI and it is taught what to do / when before getting into player end controls.
Which is why the type of commands I stated would be the minimal type necessary for providing a bit for flexibility of preferred playstyle and general tactics.

If we can't get heuristic AI working for pets and have to rely on more traditional methods more "fine tuned" controls might be necessary, but considering we can build the NPC / player pet logic from the outset of design instead of having to work with older code to make something new happen, we can still probably avoid such problems.
More of a wait and see situation and any further thought falls into the realm of purely speculative. But keep the ideas coming as we don't know the full extent of what exactly will happen.

Should Segev's mad scientist efforts prove fruitful, we are all in for one wild ride : D


I don't use a nerf bat, I have a magic crowbar!
- Combat Mechanic -
Tech Team.
Lost Deep
Lost Deep's picture
Offline
Last seen: 2 years 3 months ago
11th Anniversary Badge
Joined: 03/20/2015 - 17:48
To be honest, if this system

To be honest, if this system is in the game, I'm unlikely to use it. That said, the City of Heroes trio of commands did need refinement: I wound up making separate buttons for the stances and the actions so that my pets would stop going into aggressive mode just because I told them to attack someone in particular. Little things like that would be great to avoid.

Let's consider a simpler method, using UI. My thought is 1. for people who prefer the UI over keybinds and 2. for people to easier make keybinds, instead of having to write code.

I'm written enough code to know that it can be aggrivating, even with simple code.

This hypothetical UI panel would have three rows: Pet, stance, action. Choose the pets you want to change, what action they should be doing, and their stance. Also, there would presumably be a "go" button, so that the pets don't DO anything until you're ready.

Tier 1/follow/defensive: you have your little guys on bodyguard duty.

Tier 3/go there/aggressive: you send you big guy to ruin that area.

All/Attack target/Defensive: ruin that guy's day (but don't get carried away!)

From here, people can keybind things. But, importantly, if you don't have a situation that you need on your keyboard, you're three or four clicks away from having it. While not as effective as complete keybinding, it or a similar system could permit more complete control without needing to completely fill the keyboard.

On a separate note, from a robots/poison MM, yes, some method of priority controls would be nice.

Under Construction...

Gorgon
Gorgon's picture
Offline
Last seen: 1 year 5 months ago
11th Anniversary Badge
Joined: 05/15/2014 - 11:46
I never, ever, ever needed

I never, ever, ever needed that level of command micromanagement.

I had attack, passive, goto, and follow defensive (bodyguard), directed at all pets. If I needed finer granularity on rare occasions, i just selected and issued the commands the hard way.

I had precisely one micromanagement command, where I would tell minion 1 (by name) to attack, used to pull. Tell them to attack, as soon as they shoot, passive to run back. Sometimes they made it, sometimes not. Sometimes they pulled 1, sometimes the whole group.

Pulling working as intended!

The rest of the time was a slow, plow the road bodyguard existence.

__________________

The very existence of the taunting tank irritates, for it requires idiotic AI that obeys the taunt.

planetzoo2.0
Offline
Last seen: 3 months 1 week ago
Joined: 01/06/2021 - 09:57
hey, i just discovered this

hey, i just discovered this game a few weeks ago--so correct me if i say something dumb--but how about having a single alpha pet/minion. and that alpha has like, a radius of how far away the pets wander from it, that you can change, so that you can reduce unwanted aggro. and another potential idea, along with attacking, remaining passive, protecting you, staying still, following, and going somewhere, how about 2 or 3 attack formations. for some 2000 iq gameplay.
.

Foradain
Foradain's picture
Offline
Last seen: 3 days 8 hours ago
kickstarter11th Anniversary Badge
Joined: 10/25/2013 - 21:06
Welcome Aboard, planetzoo2.0!

Welcome Aboard, planetzoo2.0!

I don't know what the plans are for the initial options when they release the Commander archetype, but I do know that it won't be at launch, barring unforeseen events. But at the least I think these are ideas worth considering for future updates of the Commander.

Foradain, Mage of Phoenix Rising.
Foradain's Character Conclave
Millennium City Refugee.
Avatar courtesy of Satellite9

planetzoo2.0
Offline
Last seen: 3 months 1 week ago
Joined: 01/06/2021 - 09:57
would they be at all easy to

would they be at all easy to code?

Lothic
Lothic's picture
Offline
Last seen: 8 hours 45 min ago
kickstarter11th Anniversary Badge
Joined: 10/02/2013 - 00:27
planetzoo2.0 wrote:
planetzoo2.0 wrote:

would they be at all easy to code?

Unfortunately like Foradain said the Commander archetype is NOT going to be available when the game first launches so it's probably safe to assume these things are NOT going to be easy to code. Even the original "pet commander" class (the Mastermind) from The Old Game (TOG) was not released until 18 months after the original game launched.

CoH player from April 25, 2004 to November 30, 2012

.Foresight
.Foresight's picture
Offline
Last seen: 1 day 12 hours ago
Joined: 11/24/2014 - 10:48
I just joined the City of

I just joined the City of Titans discord yesterday and it looks like about 6 months ago they discussed possibly being able to do a Dragon Age style A.I. customization. I've dreamed of a day when mmo pets could have that kind of control, since City of Villains was my first MMO. I used to macro extensively to have elaborate control of my pets based on my circumstances, but none of it was inherent to it's A.I. aside from including passive/aggressive/defensive in the macro along with whatever other /petcom.

With a Dragon Age type one, you could prioritize actions based on various combat scenarios. For example, changing A.I. from offensive to defensive at certain player health thresholds and have the pet cast its buff or heal ability, which could then shift it back to its offensive programming if that pushes you above the designated health threshold.

"My name is Ozymandias, King of Kings: Look on my Works, ye Mighty, and despair!"

Airhead
Airhead's picture
Offline
Last seen: 3 hours 32 min ago
kickstarter11th Anniversary Badge
Joined: 10/03/2013 - 23:38
Neural networks (assuming

Neural networks (assuming they can be configured performantly) have the potential to derive any number of actions based on any number of situational metrics. They could be retrained, offline or real time, to train them smarter or dumber. I fear many players might find such unpredictability too painful. Other players might find loopholes in behaviour the developers could not predict, given the black-box nature of neural networks. Still, probably worth trying if you have a willing code nursemaid. Sounds exciting to me.

"The illusion which exalts us is dearer to us than ten thousand truths." - Pushkin
"One piece of flair is all I need." - Sister Silicon

warlocc
warlocc's picture
Offline
Last seen: 8 hours 31 min ago
Developerkickstarter
Joined: 09/20/2013 - 16:38
.Foresight wrote:
.Foresight wrote:

I just joined the City of Titans discord yesterday and it looks like about 6 months ago they discussed possibly being able to do a Dragon Age style A.I. customization. I've dreamed of a day when mmo pets could have that kind of control, since City of Villains was my first MMO. I used to macro extensively to have elaborate control of my pets based on my circumstances, but none of it was inherent to it's A.I. aside from including passive/aggressive/defensive in the macro along with whatever other /petcom.

With a Dragon Age type one, you could prioritize actions based on various combat scenarios. For example, changing A.I. from offensive to defensive at certain player health thresholds and have the pet cast its buff or heal ability, which could then shift it back to its offensive programming if that pushes you above the designated health threshold.

I've brought this up internally a couple times. I've been told basically that Dragon Age level control is possible, but without someone to spend real significant time on it, the only games that are going to do that are ones that make it a core of the game. Hard to devote the kind of time and effort to it when there's so much else to do.

That said, we have had good response on being able to write our own macros and perhaps even scripts, so that's what I'm looking forward to.

Who knows though, the future is unpredictable.

PR Team, Forum Moderator, Live Response Team

Huckleberry
Huckleberry's picture
Offline
Last seen: 5 hours 33 min ago
Joined: 01/03/2016 - 08:39
warlocc wrote:
warlocc wrote:

That said, we have had good response on being able to write our own macros and perhaps even scripts, so that's what I'm looking forward to.

Who knows though, the future is unpredictable.

I think this sounds intriguing.

We know that MWM has promised that CoT is designed to have a very low skill bar for entry. And we know that it is also going to have accessibility features for people who can't twitch as well as others. So there must also be an upper limit to the affect that skills and abilities can actually affect gameplay in order to keep that promise.

Having said that, the issue of customizing macros and perhaps even scripts is a facet worth considering.

In this thread it seems we are only talking about Commanders' minions, so I will be clear that is all I'm talking about as well.

Tannim222 has said that our current plans for minion behavior is to give them modules like NPCs have, or something like it at least. So this thread, combined with that knowledge leads me to the following thought:
Perhaps MWM can create a NPC module behavior tool with a user-friendly UI. Make it something the actual game designers use so they can program all the NPC behaviors in the game. Then when the Commander archetype is released, make it (or a version of it) available for the players so they can edit the behavior of their commander minions. Obviously commander minions would come complete with default modules; but with the behavior editor, players can optimize their minions for their own play style or even save some setups for specific scenarios. But it doesn't stop there. I can imagine a community forum where commanders can share their favorite behavior modules and discuss and critique them. The end result would be some players with the talent and time to develop great minion modules and other players who, without the time or talent, would still be able to incorporate them. In other words, we can keep the very low skill bar for entry and at the same time accommodate those with higher skill and ability. Furthermore it also ends up crowdsourcing the fine tuning of minion AI development. You know, that last 2% fine tuning and debugging that takes twice as long as the previous 98% combined. This means we could get Commanders and their minions out to beta testing sooner rather than later.
Another use for the NPC behavioral module tool is that when a Mission Architect is released, it could also give players the ability to modify the behavior of their own NPCs in the missions they create.

I know the first thing that some people will object to is the complexity and potential for abuse and potential for failure. True. But with some hard-coded limits and safety features, I don't see why it wouldn't be a fun feature. Besides, who doubts there will be a few players who truly master it and share their wisdom with the community? There will always be those helpful players who take pride in their craft as parsers of digital arcana. There always have been and there always will be.


I like to take your ideas and supersize them. This isn't criticism, it is flattery. I come with nothing but good will and a spirit of team-building. If you take what I write any other way, that is probably just because I wasn't very clear.
Tannim222
Tannim222's picture
Offline
Last seen: 5 hours 6 min ago
Developer11th Anniversary Badge
Joined: 01/16/2013 - 12:47
Huckleberry wrote:
Huckleberry wrote:
warlocc wrote:

That said, we have had good response on being able to write our own macros and perhaps even scripts, so that's what I'm looking forward to.

Who knows though, the future is unpredictable.

I think this sounds intriguing.

We know that MWM has promised that CoT is designed to have a very low skill bar for entry. And we know that it is also going to have accessibility features for people who can't twitch as well as others. So there must also be an upper limit to the affect that skills and abilities can actually affect gameplay in order to keep that promise.

Having said that, the issue of customizing macros and perhaps even scripts is a facet worth considering.

In this thread it seems we are only talking about Commanders' minions, so I will be clear that is all I'm talking about as well.

Tannim222 has said that our current plans for minion behavior is to give them modules like NPCs have, or something like it at least. So this thread, combined with that knowledge leads me to the following thought:
Perhaps MWM can create a NPC module behavior tool with a user-friendly UI. Make it something the actual game designers use so they can program all the NPC behaviors in the game. Then when the Commander archetype is released, make it (or a version of it) available for the players so they can edit the behavior of their commander minions. Obviously commander minions would come complete with default modules; but with the behavior editor, players can optimize their minions for their own play style or even save some setups for specific scenarios. But it doesn't stop there. I can imagine a community forum where commanders can share their favorite behavior modules and discuss and critique them. The end result would be some players with the talent and time to develop great minion modules and other players who, without the time or talent, would still be able to incorporate them. In other words, we can keep the very low skill bar for entry and at the same time accommodate those with higher skill and ability. Furthermore it also ends up crowdsourcing the fine tuning of minion AI development. You know, that last 2% fine tuning and debugging that takes twice as long as the previous 98% combined. This means we could get Commanders and their minions out to beta testing sooner rather than later.
Another use for the NPC behavioral module tool is that when a Mission Architect is released, it could also give players the ability to modify the behavior of their own NPCs in the missions they create.

I know the first thing that some people will object to is the complexity and potential for abuse and potential for failure. True. But with some hard-coded limits and safety features, I don't see why it wouldn't be a fun feature. Besides, who doubts there will be a few players who truly master it and share their wisdom with the community? There will always be those helpful players who take pride in their craft as parsers of digital arcana. There always have been and there always will be.

The modules govern basic behaviors. Our traffic AI for vehicles on the road is a module.

Our NOCs for combat currently have melee and ranged modules. This governs what types of powers they use for combat based on their range to the enemy.

From there, how the powers function are based on the parameters we apply. Stuff like activate on X momentum, use X momentum, use on X health.

I doubt we will allow for modifying these attributes directly through scripting. If we do allow for modifying power functions it will be through the Aug / Ref system where day if you have a Proc that generated an effect on X value of (attribute such as Momentum / Health).

There are certainly things we want to provide as a base level of play that can be expanded upon for those who want more fine-tuned command of their pets.

What those exact functions are and how they will be accessed hasn’t been determined yet. The more complexity we add through more advanced options, the longer it will take us to implement the Archetype.

For example, one of the early concepts we had was to design the Commander pets with multiple “build types”. That is, say a ranged pet may have ranged offense and ranged support or ranged offense and ranged control.

Then the player could customize and have the pet prefer one type of powers over others. Taking the ranged offense and support options, if the player chose support, the pet would prefer to use support powers over its ranged damage.

The problem with that is the complexity of design to make sure each option is functional within acceptable bounds of performance. Which means tons and tons of testing of every possible combination.

It also means making sure each power itself is designed with this type of performance in mind. What good is having say a pet heal
Power that has a requirement to trigger on an ally with 25% health and have prefer to use support powers over attacks.
The AI could very well sit around doing nothing waiting until someone is at 25% health, use the heal, then since the power finally triggered, it uses ranged attacks.

Or if we added further decision trees, it could use a limited selection of ranged attacks then when someone is at 25% health, it would heal. But then, why bother with the preferences in the first place. The power functions have already determined when the pet would use the heal.

This is a super, simple example but the more complexity we build into the powers, can compound the complexity of setting preferences which can compound complexity of multiple Summons Sets with different mechanics.


I don't use a nerf bat, I have a magic crowbar!
- Combat Mechanic -
Tech Team.