Jump to content

Jester unable to hook or lock targets inside 30 miles. Something I don't understand or Jester limitation?


Aries144

Recommended Posts

1 minute ago, Karon said:

What if there was a sort of hook function for the pilot, active on the TID repeater, to tell Jester to centre the radar on a particular track, either datalinked or from the radar (RWS and TWS)? Would that work in your opinion? The radar would be still subject to all its limitations (e.g. centering on a Datalinked notching target in TWS won't make it magically appear), but most of the micromanaging would be removed.

About the first sentence, I guess it's just a case of adding a condition where callouts are limited to hostiles within X nm if the landing gear is down (I wouldn't use the flaps as a condition, tbh).

That's what I was trying to say, yeah. 👍

Basicly a way to tell Jester "Find that guy and lock him up" or "Find that guy and sort him #1, that guy over here as #2". The latter part might be a little too much, but the first part would make things a lot easier for pilots. IMHO.

Why not tie it to the flaps? You'd get rid of all the flap-abusers in a heartbeat. I hear MX is a big fan of that 🙃

  • Like 1

So ein Feuerball, JUNGE!

Link to comment
Share on other sites

10 hours ago, Karon said:
You understand the logic is a script, right? I'm talking about coding here.

I was replying to multiple posts, more than one of which used the term script.

Replying to your post specifically I was trying to address what you brought up:

Quote

Scripting them is simply not worth it because the AI does not know fundamental parameters such as the mission tasking, ROE, conditions for mission complete or failed and so on...

The point of a preplanning mode for Jester would be to give the AI context on mission tasking, RoE, success conditions, etc.

Regarding your questions, this goes back to what I said originally. AI sometimes has to cheat. As long as this doesn't come with blatant unfair advantages, it's OK for the AI to cheat to mimic human decision making/intutition.

Q1 - There are a few ways to handle this. The easiest is to make Jester all knowing when it comes to target ID, like most AI in DCS. A step up from this would to be base target ID on parameters like speed, altitude, RCS, and/or location. If we know the enemy is sending C-130's at 250 knots/15000 ft, then Jester knows to prioritize searching for radar contacts matching that information over fighters at 35000 ft and Mach 1.

Q2 - This is a bit vague, and there is no real target priority given here. One thing I can say is that I dislike Jester's tendency to laser focus on TWS tracks, even if they're miles away, when there are closer targets to engage. In general, shorter range should increase priority.

Q3 - This is similar to 1, the AI will probably have to cheat a bit and just know where/what the target is if it reappears soon enough. If it vanishes from radar for a long time, then there should probably be a reset function and Jester will then have to reattempt ID when it next appears based on speed/alt/etc. 

One thing to keep in mind is that I don't know how Jester/DCS is coded, so these are just examples. They may or may not be feasible as described, but I hope they get the idea across.

Awaiting: DCS F-15C

Win 10 i5-9600KF 4.6 GHz 64 GB RAM RTX2080Ti 11GB -- Win 7 64 i5-6600K 3.6 GHz 32 GB RAM GTX970 4GB -- A-10C, F-5E, Su-27, F-15C, F-14B, F-16C missions in User Files

 

Link to comment
Share on other sites

4 hours ago, Bremspropeller said:

That's what I was trying to say, yeah. 👍

Basicly a way to tell Jester "Find that guy and lock him up" or "Find that guy and sort him #1, that guy over here as #2". The latter part might be a little too much, but the first part would make things a lot easier for pilots. IMHO.

I'm not even going to open the can of worms that is the usage of flaps. That's something for you lads in the front. I'm more than comfortable in the backseat 😛

That being said, @IronMike is this something feasible, in your opinion?

47 minutes ago, Exorcet said:

I was replying to multiple posts, more than one of which used the term script.

Replying to your post specifically I was trying to address what you brought up:

The point of a preplanning mode for Jester would be to give the AI context on mission tasking, RoE, success conditions, etc.

Regarding your questions, this goes back to what I said originally. AI sometimes has to cheat. As long as this doesn't come with blatant unfair advantages, it's OK for the AI to cheat to mimic human decision making/intutition.

Q1 - There are a few ways to handle this. The easiest is to make Jester all knowing when it comes to target ID, like most AI in DCS. A step up from this would to be base target ID on parameters like speed, altitude, RCS, and/or location. If we know the enemy is sending C-130's at 250 knots/15000 ft, then Jester knows to prioritize searching for radar contacts matching that information over fighters at 35000 ft and Mach 1.

Q2 - This is a bit vague, and there is no real target priority given here. One thing I can say is that I dislike Jester's tendency to laser focus on TWS tracks, even if they're miles away, when there are closer targets to engage. In general, shorter range should increase priority.

Q3 - This is similar to 1, the AI will probably have to cheat a bit and just know where/what the target is if it reappears soon enough. If it vanishes from radar for a long time, then there should probably be a reset function and Jester will then have to reattempt ID when it next appears based on speed/alt/etc. 

One thing to keep in mind is that I don't know how Jester/DCS is coded, so these are just examples. They may or may not be feasible as described, but I hope they get the idea across.

Thanks for replying. Other observations:

Q1 - you almost lost me at cheating...
Who would tell you those parameters? A human controller cannot interface himself with Jester. The AWACS is in ED's hands, and I doubt they will ever implement anything like it.

Q2 - I was referring to the transport / fighters example. Again, no way even for a human RIO to sort this out besides working on the geometry.

Q3 - More cheats? Humans don't cheat, they guess based on experience and factors. This is not really something you can code easily.

Unfortunately, I disagree with your answers, and these are just common and simple scenarios. I'm afraid there is little way to mimic a human without cheating or rolling dice (although, in a sense, humans ponder options and do roll dice). I think @Bremspropeller's point is much more feasible tbh.
However, HB's working on Jester 2.0, and I'm looking forward to being surprised by the final implementation 🙂

  • Like 2
full_tiny.pngfull_tiny.png
full_tiny.png

"Cogito, ergo RIO"
Virtual Backseaters Volume I: F-14 Radar Intercept Officer - Fifth Public Draft
Virtual Backseaters Volume II: F-4E Weapon Systems Officer - Internal Draft WIP

Phantom Phamiliarisation Video Series | F-4E/F-14 Kneeboard Pack

Link to comment
Share on other sites

24 minutes ago, Karon said:

 

Q1 - you almost lost me at cheating...

Sadly it's the nature of AI. We can't give the AI PFM flight models, nor can we code artificial brains that 100% mimic human behavior. The AI gets simple models that take shortcuts. It's the nature of simulation.

24 minutes ago, Karon said:

Who would tell you those parameters? A human controller cannot interface himself with Jester. The AWACS is in ED's hands, and I doubt they will ever implement anything like it.

This is what the preplanning option is for. You (the person flying the mission in DCS) align Jester's priorities with the mission breifing before you takeoff.

24 minutes ago, Karon said:

Q2 - I was referring to the transport / fighters example. Again, no way even for a human RIO to sort this out besides working on the geometry.

Again we need a priority system here. If the targets are the transports, then they get priority unless the other contacts are a threat. Jester would ideally not focus the radar on one group over another without a reason (ie attacking the mission objective or self defense).

24 minutes ago, Karon said:

Q3 - More cheats? Humans don't cheat, they guess based on experience and factors. This is not really something you can code easily.

Humans don't cheat, but AI must. If they didn't we would just be flying by ourselves. If you don't like the word cheat, think of it as simplification, shortcut, or modeling. We can't model a human brain, only lines of code that externally look sort of like a human actor. You're right that guessing and experience is not coded* because it's hard, but it's also unnecessary to copy 1 for 1.

 

*If you really wanted to make the best RIO possible maybe you could try using learning algorithms to come up with an AI program. This might become really good at emulating human behavior, but it still needs to run on our computers and fit within developer budgets.

24 minutes ago, Karon said:

Unfortunately, I disagree with your answers, and these are just common and simple scenarios. I'm afraid there is little way to mimic a human without cheating or rolling dice

It's fine, not everyone agrees on everything, though I'm completely on board with what you said. However the cheating and dice rolling isn't a problem when done right. In fact, it's the only option we have.

Awaiting: DCS F-15C

Win 10 i5-9600KF 4.6 GHz 64 GB RAM RTX2080Ti 11GB -- Win 7 64 i5-6600K 3.6 GHz 32 GB RAM GTX970 4GB -- A-10C, F-5E, Su-27, F-15C, F-14B, F-16C missions in User Files

 

Link to comment
Share on other sites

I see your points @Exorcet, disagreeing is perfectly fine 🙂

My only "concern" (for the lack of a better word), is how much effort would this require from Heatblur. I have the feeling a less-cheaty-more-realistic implementation requires a series of structural changes that only ED can make. Perhaps a common framework accessible via API.
At the end of the day, the number of multicrew aircraft is growing fast, and a solution provided from ED would help every third party (but the discussion should happen in a different channel).

full_tiny.pngfull_tiny.png
full_tiny.png

"Cogito, ergo RIO"
Virtual Backseaters Volume I: F-14 Radar Intercept Officer - Fifth Public Draft
Virtual Backseaters Volume II: F-4E Weapon Systems Officer - Internal Draft WIP

Phantom Phamiliarisation Video Series | F-4E/F-14 Kneeboard Pack

Link to comment
Share on other sites

I think what we want from Jester is really simple. We need him to prioritise radar picture on closest TID DL hostile contacts within 40 miles of us as we march down the timeline.

Most of the time there is only one variable to fine tune, and that is the antenna scan elevation. 

On balance, it is not a huge problem. Tactically, you should try to co-alt your priority targets to reduce your RIO workload anyway. Provided that is tactically available option.


Edited by Zaphael
  • Like 2
Link to comment
Share on other sites

2 hours ago, Zaphael said:

I think what we want from Jester is really simple. We need him to prioritise radar picture on closest TID DL hostile contacts within 40 miles of us as we march down the timeline.

Most of the time there is only one variable to fine tune, and that is the antenna scan elevation. 

On balance, it is not a huge problem. Tactically, you should try to co-alt your priority targets to reduce your RIO workload anyway. Provided that is tactically available option.

 

Agreed. That alone would be a huge quality of life improvement. 

Link to comment
Share on other sites

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...