I've been thinking a lot about taking a slightly different approach to the RTS interface lately, with the goal of coming up with something that gives the player a high degree of flexibility, whilst being scalable enough to cope with both small and large numbers of units. It also has to be accessible, and maintain the feeling of control for the player. Anyway, the other day this idea popped into my head - the idea of strategic markers. Strategic markers are objects that can be placed by the player anywhere on the map, or even attached to units or buildings. These objects act as hints to the unit AI which will cause it to act in a certain way - meaning that with careful placement of markers and units, the player should be able to rely on the AI to do more or less what he wants it to, without having to manage every single unit at the same time - although he is free to manage any/every unit if he wants to. In some cases, units can be assigned directly to a marker using the standard point and click interface, which gives them a primary mission - this mission takes priority over any other markers nearby. You can create and destroy markers at any time, and you can only see your own markers on the map. There can be different types of marker. Here's a few basic ones to get started: Objective Marker: Some of these maybe hardcoded into the level, but the player is free to define others as he sees necessary. There are three types of objective marker: Kill, Capture, and Protect. These can be assigned to any terrain, unit or building. Area Denial Marker: This marks an area of the map that you don't want the enemy to enter, but don't necessarily wish to occupy yourself. Units will use suppressive fire on any enemy units attempting to cross this section of the map. Artillery is particularly effective for area denial, and can defend areas well out of their line of sight by continually bombarding them. Exclusion Marker: This marks an area of the map which you really don't want your men to enter, for whatever reason. The pathfinding AI will avoid these areas like the plague. Landing Marker: This marks a landing zone for dropships with reinforcements. The type of reinforcements you can drop in depends on what terrain the marker is on, and whether the marker is considered secure or not (i.e is it reasonably safe to land a big transport and roll off a few tanks there, or is it going to be a quick infantry drop and run?) So: What do people think about this approach? Any comments or improvements? Any additional ideas for marker types?

A similar thing is implemented in http://www.ysagoon.com/glob2/

That's an interesting idea.

I would be afraid of handing over too much responsibility to the AI however, resulting in a "set 'em up and just watch" sort of gameplay.

A lot of the things you are talking about are already handled by most RTS interfaces. Area Denial for example. Most games that have artillery units have an 'Attack Ground' command, so that they will just constantly bombard an area until told to stop.
I'm not sure I understand the value of a landing marker. In games that provide transports, you already have to assign the point at which you would like them to unload troops. I can't think of an instance where I'd want to unload at the same place every time, unless I'm amassing a force in a particular area just prior to attacking. But even then, by the time I am waiting for units to build and enter the transport, it's hardly any extra effort to click on the point I want them to unload.
The area exclusion marker is a good idea, particularly when the pathfinding algorithm always seems to want to take the slowest possible route somewhere.

I think it's a good idea, but perhaps it would be wise to cap the number of markers that could be placed. Either by marker type, or an overall total. It seems like if you allowed infinite used of them, the game becomes more about setting markers than actually manipulating troops yourself. Also, troops acting according to automated commands like that should NEVER be as effective as troops under explicit player-assigned orders. Maybe use a variable for discipline level which would give a random chance of troops not following orders, which is higher(or even 100%) if the player specifically gave an order (the general is watching, we'd better behave), and lower if the troops are getting orders from markers (while the cat's away, the mice will play).

I always thought the first Dark Reign game had an interesting approaching to handling AI, and it was an idea that could be improved upon. For moving around the map, it used waypoints, which could be tagged in a linked list manner to set up patrol routes, or provide explicit paths through certain areas. It also provided some basic AI strategies like: Scout, Harass, Seek & Destroy, as well as sliders for adjusting how aggressive, how soon it would seek some sort of healing/ammo, etc.

It was good since it provided enough so that you didn't have to babysit every unit, but you still had to work when things got hairy. Some improvements I would have liked were: ability to tell units to prioritize certain types of targets over others. Like attack vehicles before infantry, or buildings before mobile units. That would eliminate my tank hunters from attacking infantry (largely worthless) when there are also tanks attacking me, for example. I'd also like the ability to save "presets" once you have tweaked behavior to your liking. You can adjust all your settings for how you would like AI to behave, then save it, in order to reassign the same settings to new units.

This is a good topic.

Would players enjoy downloading/writing scripts to control their unit's behaviors?

Quote:

Original post by Trap A similar thing is implemented in http://www.ysagoon.com/glob2/

Thanks for the link, I'll have to check that out once I've sorted out a decent home internet connection.

Quote:

Original post by vHaB I would be afraid of handing over too much responsibility to the AI however, resulting in a "set 'em up and just watch" sort of gameplay.

Your initial setup would have to be impossibly amazing in order to cope with everything that your opponent might do, so I would hope that you'd only be able to get away with this if your opponent is a complete cabbage. Even then, the intention of this system is not to let the AI play the game for you, but to make it easier for the AI to do what the player wants it to do.

Quote:

A lot of the things you are talking about are already handled by most RTS interfaces. Area Denial for example. Most games that have artillery units have an 'Attack Ground' command, so that they will just constantly bombard an area until told to stop.

True, but attack ground is not quite the same as the area denial flag. The area denial flag says to all units "If you're within range of me, and you've got nothing better to do, please bombard me" whenever an enemy unit enters it. Continual bombardment is only desirable if you've no way of telling when an enemy unit enters. Furthermore, if I go and tell the artillery to go and splat some enemy unit, it can go back to bombardment of the area denial flag once it's finished, whereas with attack ground I have to specifically tell it which lump of dirt I want it to pound.

Finally, the area denial flag could double as a kind of special case exclusion flag, which will prevent units from blundering into it while it's being pounded.

Quote:

I'm not sure I understand the value of a landing marker. In games that provide transports, you already have to assign the point at which you would like them to unload troops. I can't think of an instance where I'd want to unload at the same place every time, unless I'm amassing a force in a particular area just prior to attacking. But even then, by the time I am waiting for units to build and enter the transport, it's hardly any extra effort to click on the point I want them to unload.

This isn't for moving troops about on the battlefield, it's for getting them onto the battlefield in the first place. Kind of like Ground Control's landing zones, but player defineable, and with certain limits on the type of unit you can land based on where the flag is, and how safe it is. You're right that it's a bit of an odd one out though - as far as the AI is concerned it's probably identical to a special case objective marker. It's only really seperate for interface reasons, although I suppose I might rethink that.

Quote:

The area exclusion marker is a good idea, particularly when the pathfinding algorithm always seems to want to take the slowest possible route somewhere. I always thought the first Dark Reign game had an interesting approaching to handling AI, and it was an idea that could be improved upon. For moving around the map, it used waypoints, which could be tagged in a linked list manner to set up patrol routes, or provide explicit paths through certain areas. It also provided some basic AI strategies like: Scout, Harass, Seek & Destroy, as well as sliders for adjusting how aggressive, how soon it would seek some sort of healing/ammo, etc.

Or failing that, it takes the most direct route which happens to be right through the middle of an large swarm of enemies :).

Waypoints are all very well, but it can be a pain in the arse to set up lots of subtly different sets of waypoints for a large number of units who aren't all supposed to be taking the same route. Waypoints have their uses - but I'd also quite like a quicker system where you can set up no-go zones and get your units to travel to different destinations along sensible paths with just a single click.

It might also be worth having a kind of inverse exclusion marker, which works as an attractor, so you can encourage units to pass through certain areas. For example, laying them along the bottom of a concealed wooded valley will cause units to prefer taking routes along the bottom, keeping them out of sight.

Quote:

I think it's a good idea, but perhaps it would be wise to cap the number of markers that could be placed. Either by marker type, or an overall total.

Yes, there would probably need to be some kind of upper limit, for performance reasons if nothing else.

Quote:

It seems like if you allowed infinite used of them, the game becomes more about setting markers than actually manipulating troops yourself. Also, troops acting according to automated commands like that should NEVER be as effective as troops under explicit player-assigned orders. Maybe use a variable for discipline level which would give a random chance of troops not following orders, which is higher(or even 100%) if the player specifically gave an order (the general is watching, we'd better behave), and lower if the troops are getting orders from markers (while the cat's away, the mice will play).

I'm not sure that's even necessary. I think the only way to get units to do *exactly* what you want them to do is to explicitly tell them what to do, and it probably always will be. However, in most RTS games these days you really need to watch every unit, because the unattended AI is only really able to make very basic decisions like choosing between "Stand around and do nothing", "Stand around and shoot at anything that comes near" or "Go running off into the middle of the enemy and get myself killed" depending on what AI mode you have selected for it. The markers allow the AI to make a slightly more educated guess about what you might want them to do, without having a massively complicated interface.

Quote:

Original post by pTymN Would players enjoy downloading/writing scripts to control their unit's behaviors?

Some might, but others might be put off by it. I'm deliberately steering clear of this sort of thing because I want the interface to be as easy to use as possible to anybody, and because I want it to fit in with the game world. Scriptable unit AI might make an interesting premise for a game though.

Quote:

Original post by Sandman I've been thinking a lot about taking a slightly different approach to the RTS interface lately, with the goal of coming up with something that gives the player a high degree of flexibility, whilst being scalable enough to cope with both small and large numbers of units. It also has to be accessible, and maintain the feeling of control for the player. Anyway, the other day this idea popped into my head - the idea of strategic markers.

From your examples I'll assume that markers can refer to both entities (units, buildings, what have you) and areas on the map. In the case of areas you'll want to be able to define the area, either by some sort of polygonal area or maybe (in a tile-based system) like you would define the area you want to dig in Dungeon Keeper.

Quote:

So: What do people think about this approach? Any comments or improvements? Any additional ideas for marker types?

I'm all for this system. In my opinion this system would at least be the first step in order to actually put the s in the rts, as way too many rts games are actually too focused on individual soldiers and arcade action. For crying out loud, the troops are usually too stupid to seek cover from fire automatically when fired upon.

If you want to have resource collecting, you could have harvesting markers with additional information regarding the aggressiveness of the gathering, eg. you might want to maximize the regrowth of the resource (assuming it grows back), maximize efficiency (most profit in the least time), or maybe you just want to clear the area (profit not as important as clearing the resource).

If you want to be able to build buildings, you could have different zones à la Simcity — in residential zones the peasants would build houses, markets etc.; in military zones the peasants would build barracks etc. You could configure each zone according to what you wanted to build, but the actual building placement could be automated.

Also you could have more complex markers, such as marking an area to be defended so that walls and guard towers are automatically built at the borders, patrols would be organized etc.

You could be able to prioritize the markers so that an area marked with the exclusion marker could still be entered if the current job of the troops in question has greater priority.

You could group the markers and units so that certain groups of units would seek only their corresponding groups of markers.

Planning with markers: you could have the ability to make markers become active at the moment the jobs related to some other markers have been finished; eg. team A will not move on to marker X until team B has reached destination at marker Y. Also, there could be other triggers, eg. having a certain number of resources/soldiers/peasants/whatever, all sort of delays, or by having the player trigger the event. This is really something I'd like — you could plan the whole attack first, assing different groups of units to different markers, then just give one attack command and watch as the plan is executed, taking over control only to fine-tune. Now that would be strategy. In most rts games I have trouble attacking the enemy from three directions at once, as I have to give (at least) three separate orders. This can wreak havoc on my timing.

As for the AI of the individual soldiers, you could have different modes, eg. find jobs of highest priority, assing to marker group X, do nothing unless specifically told, do nothing unless specifically told except take cover in case of battle, etc.

Quote:

Original post by vHaB I think it's a good idea, but perhaps it would be wise to cap the number of markers that could be placed. Either by marker type, or an overall total. It seems like if you allowed infinite used of them, the game becomes more about setting markers than actually manipulating troops yourself.

But isn't the whole point of strategy games planning ahead on the large scale, setting objectives for troops, not hassling with individual soldiers with "walk here"-type of commands?

Quote:

Also, troops acting according to automated commands like that should NEVER be as effective as troops under explicit player-assigned orders.

Why, why, O why? [crying]

...as in to express a deep, mind-numbing angst, since the computer player can give all orders explicitly and personally, for all 2000000000+ units every frame (well, maybe not that many, but you get the point). The player is simply not capable of giving different orders to different troops using the current method of control.

Quote:

Maybe use a variable for discipline level which would give a random chance of troops not following orders, which is higher(or even 100%) if the player specifically gave an order (the general is watching, we'd better behave), and lower if the troops are getting orders from markers (while the cat's away, the mice will play).

It's a change in interface, not in game mechanics (well, this is at least how I'd see it). If the player put a marker on the map, he effectively gave an order. The general doesn't give the orders personally, he gives his subordinates the order to give the orders to their subordinates etc. ...ad nauseaum, or until the troops finally receive the orders. Giving and order "personally", as the traditional "selecting a unit and giving a personal order", is not giving an order personally proper — after all, the player is practically omnipresent in these games anyway, but the general, being the player's physical manifestation in the game world, could not be. If you truly want to model something like this, then at least do it properly: have officers of different ranks actually leading the troops, and give the troops a moral boost or something. The rts genre has become too close up and personal as it is, focusing on individuals (heroes and what have you) instead of relying on collaboration of large amounts of troops fighting together.

Don't get me wrong, I do like the idea of moral, but as in rpgs, the player is not an entity in the game world per se — if you need a manifestation of the player in the game world, then you need to apply the same rules to this manifestation than any other character. As far as I know, most generals aren't omnipresent, omniscient, or omnipotent. The control mechanism in rtss is currently horrible. Having the player to optimize every job assignment by giving the jobs personally would be a boring chore.

I like the idea personally and think with some work it could be an excellent new move for RTS games. But rather then just placing markers, what if instead there was a strategic map? On the strategic map the player could mark up with diffrent kinds of strategic marks?

For instance you could draw a line on the map with defense marker to create a defensive line that friendly units would attempt to hold that line against attackers. You could also set a reinforcement area and attach it to the defensive line which would cause any free troops in that area to rally at the defensive line should an attack occur. Alert areas could be used to send alert if an enemy unit is sighted in the area. Areas of the map could marked off as avoid to cause troops to avoid them. Prefered route areas could be assigned which would have a higher rating in the pathfinder. Harassment Areas would cause units in range to attack enemies in the area but not purse them out of it.

You could also set objectives in the strategic map such as buildings/units to destroy, avoid, or capture. And so.

Its an idea with alot of potential.

Quote:

Original post by Grim From your examples I'll assume that markers can refer to both entities (units, buildings, what have you) and areas on the map. In the case of areas you'll want to be able to define the area, either by some sort of polygonal area or maybe (in a tile-based system) like you would define the area you want to dig in Dungeon Keeper.

Initially I was thinking along the lines of simply dropping a marker and it affecting a circular area of variable size, mainly because it's quick and simple. You could build up more complex shapes with multiple markers of the same type.

However, I really like Technogoth's idea of scrawling directly on the strategic map... I'm going to think about this in a bit more detail and expand on it.

Quote:

I'm all for this system. In my opinion this system would at least be the first step in order to actually put the s in the rts, as way too many rts games are actually too focused on individual soldiers and arcade action. For crying out loud, the troops are usually too stupid to seek cover from fire automatically when fired upon.

Exactly. I don't think it's because the AI programmers are too crappy or the algorithms are especially complex - I think it's largely an interface issue. Most games give you some basic settings which barely even begin to cover the enormous range of behaviours you might want to see, but keep the interface simple. Some games try to give you more options, but that just balloons the complexity of the interface and makes many of those options too complicated to ever actually use, particularly in the heat of battle.

Quote:

If you want to have resource collecting, you could have harvesting markers with additional information regarding the aggressiveness of the gathering, eg. you might want to maximize the regrowth of the resource (assuming it grows back), maximize efficiency (most profit in the least time), or maybe you just want to clear the area (profit not as important as clearing the resource). If you want to be able to build buildings, you could have different zones à la Simcity — in residential zones the peasants would build houses, markets etc.; in military zones the peasants would build barracks etc. You could configure each zone according to what you wanted to build, but the actual building placement could be automated. Also you could have more complex markers, such as marking an area to be defended so that walls and guard towers are automatically built at the borders, patrols would be organized etc. You could be able to prioritize the markers so that an area marked with the exclusion marker could still be entered if the current job of the troops in question has greater priority. You could group the markers and units so that certain groups of units would seek only their corresponding groups of markers. Planning with markers: you could have the ability to make markers become active at the moment the jobs related to some other markers have been finished; eg. team A will not move on to marker X until team B has reached destination at marker Y. Also, there could be other triggers, eg. having a certain number of resources/soldiers/peasants/whatever, all sort of delays, or by having the player trigger the event. This is really something I'd like — you could plan the whole attack first, assing different groups of units to different markers, then just give one attack command and watch as the plan is executed, taking over control only to fine-tune. Now that would be strategy. In most rts games I have trouble attacking the enemy from three directions at once, as I have to give (at least) three separate orders. This can wreak havoc on my timing. As for the AI of the individual soldiers, you could have different modes, eg. find jobs of highest priority, assing to marker group X, do nothing unless specifically told, do nothing unless specifically told except take cover in case of battle, etc.

Some excellent ideas here. I particularly like the idea of group specific and triggerable markers, so that you could perhaps trigger a whole set complex manouvers with a single keypress, although I'm not sure yet how you could expose that to the interface without making it frighteningly complicated. I'll have to think about that a bit more.

Quote:

Original post by TechnoGoth I like the idea personally and think with some work it could be an excellent new move for RTS games. But rather then just placing markers, what if instead there was a strategic map? On the strategic map the player could mark up with diffrent kinds of strategic marks? For instance you could draw a line on the map with defense marker to create a defensive line that friendly units would attempt to hold that line against attackers. You could also set a reinforcement area and attach it to the defensive line which would cause any free troops in that area to rally at the defensive line should an attack occur. Alert areas could be used to send alert if an enemy unit is sighted in the area. Areas of the map could marked off as avoid to cause troops to avoid them. Prefered route areas could be assigned which would have a higher rating in the pathfinder. Harassment Areas would cause units in range to attack enemies in the area but not purse them out of it. You could also set objectives in the strategic map such as buildings/units to destroy, avoid, or capture. And so.

As I mentioned earlier in this post, I was originally thinking along the lines of simple circular areas of varying radius for the sake of simplicity, but I really like the idea of being able to scrawl troop movements and other details directly on to the strategic map.

Thanks to everyone for the feedback so far, there are some really nice ideas in here.

This thread is full of excellent ideas for all RTS in general. I know the RTS that my group is creating will benefit from it.

One thing I'd suggest is that this system should be used as a way to let all the unit AI know what the player's strategic hotspots/concerns are, not necessarily to let the AI decide the best way to go about handling them. Someone mentioned defining an area to defend and having the AI units build towers and turrets to strategically defend it. The player should still be responsible for knowing how best to defend his area and therefore knowing where to place his buildings within his area. A better instruction may be to tell units to patrol an area and stay within a certain border. Of course, having priority for different instructions/strategic markers was one of the great ideas mentioned in this thread, so if a particularly formidable attack came into the defended area, the units may have a higher priority instruction to retreat in such a scenario. Again, very good ideas here to be discussed.

Hrm, interestesting topic. Check out an old fun PC game called Majesty, (damn the developers for taking forever to make the sequel, i swear they've completely forgot about it because of the other Playerboy Mansion project.) it's an rts game that uses flag for strategic markers. And you can't control the units directly, but have to use these bounty flags, place gold to increase the attraction weight toward the npc units.

You need to address the issue of conflict of multiple markers attracting the same npc. How do you determain which marker the npc should follow, or is there some kind of mixed reaction?