[Home]GouldingiHT

Robo Home | Changes | Preferences | AllPages

A robot derived from Gouldingi by PEZ and FloodMicro by Kawigi, and probably the only robot that either of us have made that we can watch beat SandboxDT over 100 rounds occasionally (although certainly not consistently). The fact that it's still a MiniBot is partially due to Kawigi fitting a version of his SandboxFlattener movement, still basically fully-featured, into a MicroBot.

How does it move?

It uses SandboxFlattener movement, slightly improved from FloodMicro (v. 1.0) to avoid getting nailed up against the walls.

How does it fire?

PEZ wrote the gun, using AngularTargeting/Factored and AveragedBearingOffsetTargeting. The same as in Gouldingi.

Though quite much shrinked by both me and Kawigi. -- PEZ

How does it dodge bullets?

In principle it does the same thing as FloodMicro and MakoHT does, discussed on the SandboxFlattener page. Basically, it tries to give statistical aiming no better than a random chance of hitting by moving in random lateral directions and speeds when it observes a bullet being fired. However it doesn't decide to react to every bullet, and it learns which optimal distances it should stay from its enemies for the best net benefit.

How does the melee strategy differ from one-on-one strategy?

It is only designed for OneOnOne battle, so there's a good chance it would act somewhat silly in a Melee battle. I could be wrong, however.

How does it select a target to attack/avoid in melee ?

Man, it would get confused detecting bullets in Melee...

What does it save between rounds and matches?

Between rounds it saves its targetting factor and its benefits and penalties for various distances.

Where did you get the name?

Gouldingi gets its name after a type of piranha, the HT is added by its characteristic movement, and it stands for "Hit This". Coincidentially, I think Gouldingi's gun is about the best one I've seen to hit this movement.

Can I use your code?

Yes, GouldingiHT is an OpenSource bot: /Code

What's next for your robot?

If we can free up some space to keep it a MiniBot, we might add some better energy management and make it ram disabled opponents.

What other robot(s) is it based on?

Gouldingi, of course, and FloodMicro. It also came from older versions of Mako, and is a cousin to MakoHT.

Comments

Watch your radar management. It gets the opponent on the edge of the arc instead of the center, so they slip out routinely. The lost scans are killing you. -- Kuuran

That's true. PEZ took out a more consistent radar scheme in favor of Moebius's radar system to save space, so it does slip occasionally. About 50% of the time (you can't really tell by watching), it only misses one turn, but the rest of the time it misses 6 or 7 turns to get the enemy back. -- Kawigi

Fair enough. There are a few one-line schemes for OneOnOne on the Radar page posted by Albert and David Alves which you might want to take a look at, though. Shouldn't be too much space to give up for a more reliable system, but then, I haven't seen Moebius's radar, so I could be wrong :) -- Kuuran

This shouldn't be too bad I think. The guns are so simple that they still do their job pretty well even if we skip turns now and then. At the moment we only have 1 byte left so the minimized radar is kind of important. But maybe if we find more bytes to save elsewhere we can replace it with a more consistent one. Once we have released source code maybe this will be solved. -- PEZ

Ironically, FloodMicro currently outranks GoundingiHT? on the Eternal Rumble, and nearly matches MakoHT. I am pretty sure the movement is completely equivalent, except that GouldingiHT will not try to move away if it is backed against a wall (which I wouldn't think would make it less effective, except maybe on a pure circular aimer). This would seem to suggest that an Angular gun is not necessarily better than a head-on gun against many better robots. Or maybe you need to re-introduce that bug you fixed from the targeting, PEZ. -- Kawigi

The bug is still there, don't worry. =) I haven't uploaded version 1.1 of GouldingiHT yet. Wanted to see how 1.0 fared first. I think that HeadOnTargeting can prove super strong in many cases. Though I hope not it will prove better than AngularTargeting/Factored =) But maybe time will tell. I'm still confident HT movement will eventually force Paul Evans to upgrade SandboxDT. But maybe it will take an overhaul of MakoHT or some such before HT movement reaches high enough in the rankings to prove it. I'm thinking a MakoHT with both Gouldingi and HT movement, and the intelligence to choose among them, might be able to really fight among the big guys. -- PEZ


Robo Home | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited May 18, 2006 21:50 EST by Corbos (diff)
Search: