Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Suoli]

The concept:
A network of animal powered watch towers scattered around the surface with the added feature of each tower sending an appropriate signal to a central computing system when a hostile is spotted in a given direction from the tower. With simultaneous signals from three or more watch towers, it should then be a small matter of engineering to create a computing system that can combine the signals and triangulate more or less accurately the position of the hostiles.

Applications:

• There's been talk of an aimable minecart cannon lately. Depending on the maximum range and level of control on such a system, you could hook it up to an automatic targeting system and trace the footsteps of invading goblins with a rain of magma.
• If that doesn't pan out, you could just use magma mines or other more localized trap types instead.
• A dwarven war room with a miniature map of the surface level and a door/hatch based display system to show enemy positions and movements in real time. Think Dr. Strangelove. Appropriately color coded building materials for the map, levers connected to your defense mechanisms and a big table with chairs surrounding the map.

Execution:

• Watchtowers
   I can see two ways to go.
  Firstly, each of the watchtowers could consist of 4 pilbox units radiating from the center of the tower to each direction. Each unit would a field of view restricted to a given direction, one square to pen the watch animal, one square for a pressure plate and an escape path leading over a hatch that covers a pit. At the center, there would be a downward stairway that leads to safety. When the penned animal detects a hostile, it paths for the central stairway and steps on the pressure plate. The pressure plate sends a signal to the central computing system and opens the hatch, blocking the path to safety. Ideally, the animal would then stand on the pressure plate for as long as the hostiles are in sight and then return to it's pen.
   Alternatively, it might be possible to sort of invert the watchtower by placing a single animal pen in the center and 4 escape paths on the edges. This would only work if, when detecting a hostile, the animal always chose the escape path furthest away from the enemy.
• Signal processing
   Yeah, not touching this until other aspects of the project have been tested and verified. I'm pretty sure it's doable with minecart logic.

Unresolved issues and stuff that needs testing:

• Details of creature logic, i.e fleeing and pathing behaviour, what happens when an escape path is blocked, will they return to their pen on their own etc.
• Finding the optimal detector creature
• Details of the field of view mechanics
• Just about everything concerning the signal processing (low priority for now)

At this point it's mostly brainstorming and untested theories but do you think this could be viable? Either way, all suggestions, information and any form of participation would be greatly appreciated. I have very little experience or knowledge with most of the mechanics involved here and there's lots of stuff that needs to be researched before I'm going to even attempt this.

[joeclark77]

I'm not so sure "triangulation" is worth the effort.  What you might do instead is have a bunch of north-east facing emu towers that are each at the southwest corner of each map quadrant (which are approximately as tall and wide as the emu's range of vision).  Then very simply each emu turns a war-room-map tile "on" or "off".  You still get the same mini-map effect without having to do any computation.

Much more glory to you if you can make triangulation work, of course!

[Suoli]

I'm not so sure "triangulation" is worth the effort.  What you might do instead is have a bunch of north-east facing emu towers that are each at the southwest corner of each map quadrant (which are approximately as tall and wide as the emu's range of vision).  Then very simply each emu turns a war-room-map tile "on" or "off".  You still get the same mini-map effect without having to do any computation.

Much more glory to you if you can make triangulation work, of course!

Well, I could point out that triangulation is potentially more scalable and has higher definition for fewer emus, but to be honest... it's the effort itself that's really worth the effort for me. Having to figure out computation is a bonus, not a negative.

[Girlinhat]

Do animals no longer generate instant "enemy spotted" alerts?

[Mr S]

Using a computer, to emulate emus, to emulate a computer.

Win.

I love Bay12.

[Suoli]

Do animals no longer generate instant "enemy spotted" alerts?

They do, but what they can't do is order artillery strikes on their own. Yet.

[PrimusRibbus]

Emu Radar... lmao. Never change, DF!

[fricy]

What a dwarfy idea!
I'm not sure how pastured animals will behave when spotting the enemies, and if they will return to their window on their own. I suspect dwarf power will be needed to reset the system.
You'll need to force the pathing of invaders for efficient aiming, altough if you combine retracting bridges with derail ramps you should be able to make a cannon that has variable range. You need to hit a wall though if you want to spill magma/water/spears out of the car, and not just hit them with the cart itself. Building choke points with animal sentry at the entrance sounds good anyway, you can use them to start a repeater that operates spikes too.

[Ace_Warbringer]

As far as I can tell any non-war animal runs from nearby danger (which may need a path both for the attacker or the attackee.) then generates a haul animal to pasture job. which my dwarfs must immediately drop everything to do.
I often have a general pen near my entrance for grazers early on witch doubles as early warning for thieves and 'nappers.

so i'm not sure if an animal watchtower would help.

[slothen]

the grid system would make any triangulation system of poor resolution.  If 3 wachtowers go off at once, at best you'll know the enemy is roughly within the triangle, but you won't know its true position.  My suggestion would be to reduce it to one dimension.  That is, make a series of East West roads/paths to your fortress, then implement such a system that selectively activates traps/defenses on that road alone, using signals from the North and South end of that road.  This would be both doable within the confines of the game, and stepping stone to more ambitious projects, such as pursuing actual triangulation.  you can put additional towers to monitor enemy progress on the road, setting off more events.

[Suoli]

As far as I can tell any non-war animal runs from nearby danger (which may need a path both for the attacker or the attackee.) then generates a haul animal to pasture job. which my dwarfs must immediately drop everything to do.
I often have a general pen near my entrance for grazers early on witch doubles as early warning for thieves and 'nappers.

so i'm not sure if an animal watchtower would help.

My first plan didn't go too well but I just had some success with the following test setup, using a meeting zone instead of a pen:

Code: [Select]

#########
#gImh^H.#
#########


Code: [Select]

g - Goblin
I - Fortification
m - Meeting zone
^ - Pressure plate linked to the hatch
h - Hatch
H - Giant hamster
. - Floor
Things to note:
* There's no actual escape path that leads out of hostile LOS.
* There's no path between the goblin and the hamster.

First, I drop the hamster in from above. After recovering from the stun, it paths towards the meeting zone and triggers the plate. Hatch opens, blocking the path, and hamster keeps standing there like a moron. This is why I was having trouble with my other plan: when the hamster's path was suddenly blocked, instead of pathing somewhere else it just stood there indefinitely. In this design, the weird pathing thing is used to keep the hamster in place when there's no hostiles in sight.

Next, I drop the goblin on the other side of the fortifications. At this point, the comatose hamster goes haywire and starts running back and forth between the pressure plate and the other end of the corridor. It was being a bit of a jerk up until that point so seeing it panic was most satisfying. I'll probably place another pressure plate on what is currently empty floor and use that as the output.

Next up:
* Hatchless design. Is the path blocking trick necessary? Will the hamster path to the meeting zone and stay there or will it wonder around?
* Range testing. Does distance figure into the fleeing behaviour? Getting different behaviour at different ranges could be amazingly helpful for calculating position.

Hmm. I wonder if it would be possible to create angles that obstruct line of sight from a certain distance but not from another. Like so:

Code: [Select]

Sideview:
H
####
####
####   x          g


Code: [Select]

H - Giant hamster
x - Goblin out of FOV
g - Goblin within FOV

Anyone want to help test some of this stuff? It seems like useful information for other projects as well.

[Oaktree]

An alternate more "point" detection is putting the animal overhead sitting on top of a grate.  Might not get the same detect ranges, but leaves open firing lines if shooting things horizontally or flooding an area.

[Suoli]

the grid system would make any triangulation system of poor resolution.  If 3 wachtowers go off at once, at best you'll know the enemy is roughly within the triangle, but you won't know its true position.

If I end up doing a grid where each node has line of sight to every direction I will space them so that their spheres of detection have only partial overlap. If the detector has a vision radius of 20 and I space them 30 squares apart, then that should make the system accurate to 10 squares. Given the spread of minecart shotguns, size of goblin squads and the fact that they move around anyway, that should be good enough. Or I might think of an even more accurate system. But before even considering things like that, there's lots and lots of animal testing to be done.

[Suoli]

An alternate more "point" detection is putting the animal overhead sitting on top of a grate.  Might not get the same detect ranges, but leaves open firing lines if shooting things horizontally or flooding an area.

Or put a hatch covered pit under the grate. Wow. I don't think there's any other elegant way to trigger a hatch right when a hostile has stepped on it. Also detects kobolds while letting merchants come through. This could have huge implications for a lot of trap designs.

[wierd]

Emu Radar... lmao. Never change, DF!

It's perfect! Emus love shiny objects, and goblin invaders have all that shiny goblinite on!

Not sure how to get them to trigger an automatic artillery strike though.