Project Orbiter Battle Simulation Project Needs Developers!
- Thread starter escapetomsfate
- Start date
wehaveaproblem
One step closer
- Joined
- May 18, 2008
- Messages
- 913
- Reaction score
- 0
- Points
- 16
- Location
- London
- Website
- wehaveaproblem.wordpress.com
Project is coming along nicely, looks quite tasty!
That taxi/autopilot AI is looking particularly sharp. Nice work.
That taxi/autopilot AI is looking particularly sharp. Nice work.
RacerX
Donator
Has this project been shelved? The website in first post is no longer. Any news on this?
RisingFury
OBSP developer
No, the project hasn't been shelved. Actually, we've made a lot of behind the scenes progress, but such that we just can't show in a YouTube video. The AI is now at the point where it doesn't just think for itself, but implements squadron behavior, such as calling for help when one member is attacked,...
As for the website... the server wasn't mine. It was paid for by the leader of a UT clan that collapsed years ago. I was frankly surprised that it lasted that long. I had access to the server through FTP, but couldn't get in touch with the admin for years...
As for the website... the server wasn't mine. It was paid for by the leader of a UT clan that collapsed years ago. I was frankly surprised that it lasted that long. I had access to the server through FTP, but couldn't get in touch with the admin for years...
RacerX
Donator
Thats good to know. Im very curious if some or not all will work in OMP. I think OMP is at a point now where this might just work.
RisingFury
OBSP developer
I'm not sure if this will work with OMP. We're not even trying to make it work with OMP...
That's the point! AI so advanced that it can challenge the human pilot in both strategy and tactics.
Unfortunately with this project, it's all or nothing. There have been attempts at creating a combat system, but all have failed - the main reason is because there's never any opposition. Shooting at planes flying straight and level or bombing stationary targets on the ground can only be so much fun.
This project combines and insane amount of functionality:
- Very advanced AI and autopilot system.
- Weapon systems and weapons
- RADAR
- Damage systems with the ability to destroy vessel's systems like main thrusters, RCS thrusters, OBSP specific components like weapon system, RADAR,... as well as aerodynamics. ETF's been using his black magic in a way that it allows us to modify vessel's aerodynamics, so when a vessel is hit, it can behave as if its wing was blown off...
- Basic collision detection that works even at insane relative velocities (tested at relative velocity of 16 km/s at 100x time compression)
- Many more
But we can't release things in stages. If the AI was simply a collection of autopilots, the user would have to think for every vessel. There's no fun in that. It has to be fully built for it to make sense.
That's the point! AI so advanced that it can challenge the human pilot in both strategy and tactics.
Unfortunately with this project, it's all or nothing. There have been attempts at creating a combat system, but all have failed - the main reason is because there's never any opposition. Shooting at planes flying straight and level or bombing stationary targets on the ground can only be so much fun.
This project combines and insane amount of functionality:
- Very advanced AI and autopilot system.
- Weapon systems and weapons
- RADAR
- Damage systems with the ability to destroy vessel's systems like main thrusters, RCS thrusters, OBSP specific components like weapon system, RADAR,... as well as aerodynamics. ETF's been using his black magic in a way that it allows us to modify vessel's aerodynamics, so when a vessel is hit, it can behave as if its wing was blown off...
- Basic collision detection that works even at insane relative velocities (tested at relative velocity of 16 km/s at 100x time compression)
- Many more
But we can't release things in stages. If the AI was simply a collection of autopilots, the user would have to think for every vessel. There's no fun in that. It has to be fully built for it to make sense.
RacerX
Donator
thanks for that update. The reason I asked was now that OMP lag issues are fixed I used BombMFD in OMP and it works good. I saw some problems with it so I thought I would fire up blender and start making weapons meshes for the vessels. Then I started to think " I wonder what ever happened to OBSP?" So perhaps Ill just wait to see what you guys come up with and see what I can and can not use in OMP and go from there. Again thanks for your update
Edit
This is why I was asking before I go to far with it and using attachment points
Edit
This is why I was asking before I go to far with it and using attachment points
Last edited:
RisingFury
OBSP developer
One of the problems we've had is that both me and ETF are primarily programmers. While I've picked up some 3D and 2D skills over the years, I can't match the work that dedicated 3D and 2D artists do...
EDIT:
EDIT:
Last edited:
RacerX
Donator
Did you make those weapons? They look good. Mine are all imported and the XR1 is in there as a reference for scale and locations of future attachment points. But Im shelving this little 30 min project and wait to see what you guys come up with. And so far from what I have seen looks very impressive. :thumbup:
RisingFury
OBSP developer
These weapons were made by T.Neo.
As far as attach points... we don't use them!
In order to reduce the amount of vessels in the scenario (if every plane carried 10 weapons and if there were 20 planes, that means 220 vessels!), we spawn the weapon when it's fired. The mesh under the wings is actually added there manually in the code when the weapon is given to the plane.
ETF handled this - when you add weapons to the plane, the weapon system allocates where the weapon will go and attaches a mesh to the plane. The mesh is later removed when the weapon is fired and an actual weapon vessel is spawned.
This way, when weapons are loaded onto the plane, only the mesh shows up, but all the code that Orbiter would usually run in the vessel does not.
As far as attach points... we don't use them!
In order to reduce the amount of vessels in the scenario (if every plane carried 10 weapons and if there were 20 planes, that means 220 vessels!), we spawn the weapon when it's fired. The mesh under the wings is actually added there manually in the code when the weapon is given to the plane.
ETF handled this - when you add weapons to the plane, the weapon system allocates where the weapon will go and attaches a mesh to the plane. The mesh is later removed when the weapon is fired and an actual weapon vessel is spawned.
This way, when weapons are loaded onto the plane, only the mesh shows up, but all the code that Orbiter would usually run in the vessel does not.
Glad to hear you guys are still working on this, three years after your original announcement.
I mentioned this previously in the thread:
No offense, but I think you guys are starting from the wrong end of things, and should've started with this:
Don't get me wrong, I think you guys have accomplished a lot of cool stuff. I just think you've been putting your energies into the wrong place.
It would be significantly more interesting to have a platform with which to simulate and explore realistic-scale, realistic-physics space combat using conceivable technology. In space, you don't have to worry about terrain, so Orbiter's lack thereof isn't a problem. You also control more of what the user sees and experiences in the Orbiter engine in space, since everything that the user will conceivably interact with is a vessel that can be completely controlled by your plugin, with the Orbiter engine just doing background movement propagation.
Ground/air combat could be abstracted in the manner of a strategy game--the player doesn't need to fully experience that, because in the Orbiter engine, that experience will necessarily fall far short of what the user experiences in things like Battlefield (for infantry) or IL-2 or the DCS series (for air combat). You just need to have points of interest--a launchpad, a runway, a construction facility, whatever--something to launch missiles at or whatever.
The problem is, naturally, that no one really knows what space combat would really look like...but that's where you guys would come in, and provide something that doesn't exist currently at all, rather than yet another air combat sim that can't hope to compete with its competition.
I mentioned this previously in the thread:
It doesn't seem that I ever elaborated, so I'll do so now.
No offense, but I think you guys are starting from the wrong end of things, and should've started with this:
Ground- and air- based combat is all fine and dandy, but when it comes down to it Orbiter is a terrible platform for that sort of thing. Orbiter is first and foremost a space simulator. Adding sweet aircraft AI and sweet weapons still leaves Orbiter far behind where combat air simulators were a decade and a half ago, as far as the experience to the user goes. With no (collidable) terrain, flying a sortie through a canyon to avoid AA is kind of pointless since there's no risk of hitting the wall. With no separate water/land, there's no beach to defend since the tanks could just drive around onto the ocean. With no obstacles for cover, infantry battles are no better than Revolutionary War-era lines of musketeers shooting at each other.Space and interplanetary combat is still a long way off, but I'm confident we'll get there
Don't get me wrong, I think you guys have accomplished a lot of cool stuff. I just think you've been putting your energies into the wrong place.
It would be significantly more interesting to have a platform with which to simulate and explore realistic-scale, realistic-physics space combat using conceivable technology. In space, you don't have to worry about terrain, so Orbiter's lack thereof isn't a problem. You also control more of what the user sees and experiences in the Orbiter engine in space, since everything that the user will conceivably interact with is a vessel that can be completely controlled by your plugin, with the Orbiter engine just doing background movement propagation.
Ground/air combat could be abstracted in the manner of a strategy game--the player doesn't need to fully experience that, because in the Orbiter engine, that experience will necessarily fall far short of what the user experiences in things like Battlefield (for infantry) or IL-2 or the DCS series (for air combat). You just need to have points of interest--a launchpad, a runway, a construction facility, whatever--something to launch missiles at or whatever.
The problem is, naturally, that no one really knows what space combat would really look like...but that's where you guys would come in, and provide something that doesn't exist currently at all, rather than yet another air combat sim that can't hope to compete with its competition.
RisingFury
OBSP developer
@Hielor:
Sorry for not getting back to you sooner.
Actually, since the original announcement, OBSP has taken another turn. So far we've been concentrating on air combat and air to ground combat against vehicles. We intend to implement SAM sites and AAA, but I don't think infantry is on our table anymore. We'd like to move into space as soon as possible.
Orbital operation shouldn't be so far out. I'm currently bogged down in Squadron based AI - how do you move a squadron from one area to another, have the squadron attack the target(s) and do whatever they do afterwards. I have to take into account formation flying, allocation of units with appropriate weapons, cover,... nothing too difficult, but it is tedious. It requires two way communication between the squadron leader and all other units - but it has to be done in a way where the player can act as the squadron leader or any of the units and the AI still functions as a group.
Once I get this done and implemented, I'll be able to explode different paths - individual units already know how to approach an airfield, land, taxi to hangar or take off, but no ATC has yet been made and no patterns programmed. I think I'll do that next.
Once that's done, I have a basic atmospheric AI package done. Planes won't mindlessly fly around, they won't crash because they run out of fuel, they won't crash into each other when landing and taking off,... That'll leave me in a position where all I need to do to expand the range of the AI is to write an autopilot and then implement it into the existing AI. I can write an autopilot for orbital insertion, for plane change, for orbit sync, for docking, for reentry,... and add those to the AI. With each autopilot, the AI will know how to navigate a bit better. For example, with only orbital insertion and reentry, the AI will know how to navigate more efficiently then just flying through the atmosphere at high altitude.
This will make space combat possible. We've already collected a lot of ideas from a lot of people on how to implement it and it'll become possible to research what works in space and what doesn't.
Sorry for not getting back to you sooner.
Actually, since the original announcement, OBSP has taken another turn. So far we've been concentrating on air combat and air to ground combat against vehicles. We intend to implement SAM sites and AAA, but I don't think infantry is on our table anymore. We'd like to move into space as soon as possible.
Orbital operation shouldn't be so far out. I'm currently bogged down in Squadron based AI - how do you move a squadron from one area to another, have the squadron attack the target(s) and do whatever they do afterwards. I have to take into account formation flying, allocation of units with appropriate weapons, cover,... nothing too difficult, but it is tedious. It requires two way communication between the squadron leader and all other units - but it has to be done in a way where the player can act as the squadron leader or any of the units and the AI still functions as a group.
Once I get this done and implemented, I'll be able to explode different paths - individual units already know how to approach an airfield, land, taxi to hangar or take off, but no ATC has yet been made and no patterns programmed. I think I'll do that next.
Once that's done, I have a basic atmospheric AI package done. Planes won't mindlessly fly around, they won't crash because they run out of fuel, they won't crash into each other when landing and taking off,... That'll leave me in a position where all I need to do to expand the range of the AI is to write an autopilot and then implement it into the existing AI. I can write an autopilot for orbital insertion, for plane change, for orbit sync, for docking, for reentry,... and add those to the AI. With each autopilot, the AI will know how to navigate a bit better. For example, with only orbital insertion and reentry, the AI will know how to navigate more efficiently then just flying through the atmosphere at high altitude.
This will make space combat possible. We've already collected a lot of ideas from a lot of people on how to implement it and it'll become possible to research what works in space and what doesn't.
Pipcard
mikusingularity
spAce Combat
RisingFury
OBSP developer
In a way, yes. But real world applications unfortunately are more difficult to pull off than in Orbiter. Every time I need to estimate something in my autopilot, I take a piece of paper, account for everything and come up with a formula. That's because in Orbiter every law of Physics is exactly known, is predictable and always behaves the same way. In reality, along with limited sensors and sensor noise, you also have to deal with situations where aerodynamics doesn't always behave the same way in same circumstances.
Also, right now the my AI is programmed to do every action. In identical set of circumstances, it will always respond the same way. I haven't yet arrived to the machine learning part. A real world AI will need to learn from the enemy and disseminate that information to other units, kind of like Borg hive mind...
RisingFury
OBSP developer
Oh yes BTW, I was more thinking to an ATC addon for Orbiter
ATC *can* be completely automatic in real world, but it would be heavily reliant on sensors and technology. For example, Orbiter knows exactly where every vessel is. That gives me the ability to calculate where on a taxiway every vessel will be in the future with a decent accuracy.
I'n reality, you'd have to set up some sort of a sensor net consisting of RADAR, visual and weight sensors just to detect where planes are on the ground and how fast they're moving.
You'd also have to take into account that pilots make mistakes. They sometimes taxi faster than they're supposed to or slower than they're allowed to. They sometimes take a wrong turn or misunderstand ATC instructions. OBSP vessels won't make such mistakes. Taxi speeds are predicable, because they're coded into the autopilot. The path the vessel takes is also coded. The airfield AI calculates the path the vessel should take.
Though I will have difficulties with the human player... They're unpredictable and can cause hell for the AI
jedidia
shoemaker without legs
Then fine them! :lol:
I think what M-Nolson was refering to was that the AI could have some great application outside combat, but still inside orbiter.
Similar threads
