View source for Talk:Curve flattening

From Robowiki
Jump to navigation Jump to search

Hey, this is one of the silliest redirect page I've found ever! Will it going to be better? » Nat | Talk » 10:25, 26 April 2009 (UTC)

Contents

Thread titleRepliesLast modified
Shooting Curve Flatteners814:58, 6 August 2017
What is it?218:28, 5 June 2017

Shooting Curve Flatteners

I am testing my gun in Anti-Surfer Challenge. In the first 5 rounds I just get about %14 - %17 against Dookious. But when he opens his curve flattening. My hit rate falls to %5-%13(It changes continuously). Should a gun tuned against surfers be able to it a curve flattening bot?

Dsekercioglu (talk)08:27, 4 August 2017

I don't know how other bots solve the problem of targeting curve flatteners, but one possible solution would be to create an anti-flattener gun. Such a gun would give guess factors a negative weighing each time they were visited.

Cb (talk)11:07, 4 August 2017

I thought about that too. The problem is many bots use different slices for different segments, different algorithms. I'm not sure about that.

Dsekercioglu (talk)20:17, 4 August 2017

Yeah, maybe there is a way to detect which slices/segments/algorithms the other bots use, and then the gun could be pretty strong. I am not sure either. I guess implementing it would be complicated.

Cb (talk)20:41, 4 August 2017

Another important thing is that the bots surfs the waves in different ways. To have a %100 hit rate the bot should fire when the enemy bot is at the centre of the battlefield to remove the Wall Smoothing factor and there shouldn't be any other bullets in air(Multi-Wave Surfing).

Dsekercioglu (talk)21:15, 4 August 2017
 
 

There is a problem with this approach. Typically, you have about 10 sectors to shoot (if we divide guess factor range, by the bot body). Good curve flattener could be anywhere in a random fashion, even in just visited sector. So at best you increased your chances to 1/9 vs 1/10, in practice it is still 1/10. I have a random gun among list of available ones for my bots, often it gives the best result against curve flatteners. Sometimes, antiguess factor would lead you in wrong direction: I think DrussGT avoids GF=0, but with negative weight for visited spots, you soon will fire at GF=0 only, and unavoidably lose.

My strategy just fire randomly in this case, at least you get a chance to hit.

Beaming (talk)02:48, 5 August 2017

Actually you can increase the chance of hitting by firing when the robot is near a wall.

Dsekercioglu (talk)18:15, 5 August 2017
 

Randomly choosing an angle in the reachable area is a good fallback strategy. DrussGT does this, although this gun rarely gets used. It is more as a safety in case someone copy-pastes his gun and then surfs it.

People have experimented a lot with Anti-Surfer guns over the years, but nothing seems to work better than a GF gun that only works on very recent data. And once the enemy starts flattening there is very little you can do. At this point fighting at a distance or bulletpower where you are more competitive might be a better idea, and turn on your own curve flattening as well.

Skilgannon (talk)13:09, 6 August 2017

You are right. I just tried a weighted random gun with reverse weights and it didn't do better than the anti-surfer gun.

Dsekercioglu (talk)14:58, 6 August 2017
 
 
 
 

What is it?

You do not have permission to edit this page, for the following reasons:

  • The action you have requested is limited to users in the group: Users.
  • You must confirm your email address before editing pages. Please set and validate your email address through your user preferences.

You can view and copy the source of this page.

 

Return to Thread:Talk:Curve flattening/What is it?.

Simply said, try to design your movement so, that you have no peaks in your movement profile. WaveSurfing is designed to move to the lowest danger of your profile, but curve flattening (or just flat movement) is designed to have an equal danger on your whole range. The idea behind it is that even a good PM, GF or DC bot can't easily find a weak spot in your movement. A lot of good bots have such a movement as a secondary movement, only switching to it when they get hit too much on their primary movement. It was very popular in the period of 2004-2007, but the need for it later became much less as both targeting and movement evolved further. I know that f.e. Dookious has such a secondary movement, but I also know that you have to have a very strong gun to trigger it (GresSuffurd can't)

GrubbmGait (talk)01:05, 5 June 2017

Thank you! I didn't know that it was called curve flattening. Maybe I will write the CurveFlattening page after.

Dsekercioglu (talk)18:28, 5 June 2017