Author Topic: Autonomous Combat League  (Read 3168 times)

MarvinMan

  • Automation Expert
  • Experienced Member
  • ***
  • Posts: 236
  • Likes: +4/-0
    • View Profile
Re: Autonomous Combat League
« Reply #75 on: April 06, 2016, 11:37:48 pm »
Fair enough regarding not being able to fit any additions into the part count. I had trouble getting a bot into the middleweight class.

As unwieldy as they may be, the tail of grenade lances looks pretty cool. Have you thought about relaxing the distance/proximity requirements for detonating the grenade? It looked like there were a few cases where the grenade passed close enough to do major damage without being detonated. It could be effective to deploy all the lances in a fan pattern when space allows to give it more chances to get one of them into range.

PressureLine

  • XFM Developer
  • Administrator
  • Experienced Member
  • *****
  • Posts: 192
  • Likes: +4/-0
    • View Profile
    • Rawbots Club Forums
Re: Autonomous Combat League
« Reply #76 on: April 07, 2016, 04:47:59 am »
Not including the Legacy and Unlimited bots, we now have 11 ACL bots! (14 including the Unlimited) Well done guys!
"Never assume that anything you've never seen before is benign. Most particularly not something that says 'MARINES' on it."

PressureLine

  • XFM Developer
  • Administrator
  • Experienced Member
  • *****
  • Posts: 192
  • Likes: +4/-0
    • View Profile
    • Rawbots Club Forums
Re: Autonomous Combat League - RockSlide
« Reply #77 on: April 07, 2016, 12:41:10 pm »
New middleweight competitor!

RockSlide



Weighing in at 35 parts, RockSlide is a middleweight Plasma Shooter capable of tracking and engaging both ground based and aerial opponents.

Get RockSlide here.
"Never assume that anything you've never seen before is benign. Most particularly not something that says 'MARINES' on it."

MarvinMan

  • Automation Expert
  • Experienced Member
  • ***
  • Posts: 236
  • Likes: +4/-0
    • View Profile
Re: Autonomous Combat League
« Reply #78 on: April 07, 2016, 02:56:15 pm »
The turret on that one seem to work well, and the bot is a lot mare stable than it looks. It is a bit slower than most of the other bots, but that appears to be a good strategy in multi-bot matches as being late to the party means that it only has to deal with mopping up the survivors.

Perhaps now would be a good time to add some new variations on the basic ACL to keep things interesting. Here are a few ideas I've been thinking about:
Sumo/unarmed mode: bots with no conventional (damage causing) weapons try to flip each other or push their opponent out of the arena. Could possibly add some extra environmental hazards.
Air/spacecraft: similar to the current ACL rules, but bots capable of full 3D flight are allowed. Roughly spherical arena defined by distance to a fixed centre point.
Death race: bots try to navigate a track as fast as possible while fending off the competition. No/ limited damaging weapons. Sumo style ramming/flipping attacks encouraged.

Sleeping_Owl

  • Member
  • **
  • Posts: 18
  • Likes: +1/-0
    • View Profile
Re: Autonomous Combat League
« Reply #79 on: April 07, 2016, 07:46:34 pm »
Have you thought about relaxing the distance/proximity requirements for detonating the grenade? It looked like there were a few cases where the grenade passed close enough to do major damage without being detonated. It could be effective to deploy all the lances in a fan pattern when space allows to give it more chances to get one of them into range.
I increased detonation distance for fight with bigger bots.
The problem is that Lancer usually damages itself, and with increased distance to do it more often (detonation of almost deployed grenades), so I try to keep distance as low as possible to avoid damage to front wheels (it happened a few times). Also if target part is too close to grenade, Lancer looses target for a short time because there is some space around grenade that is not covered by cameras, these situations can be mistaken for too low detonation distance.

Deploying multiple lances should work great given enough space and some elevation control.

Quote
Perhaps now would be a good time to add some new variations on the basic ACL to keep things interesting.
Sumo - there are infinite forces, infinite masses and partially broken physics. It will need rules to control this.
Air - would be nice. I am not sure if fights in a spherical arena would be easy to watch. My zero gravity big bot construction map (a closed box) is quite unpleasant place, camera is bumping into walls and has weird direction controls.
Death race - it would be nice to get some race challenge running. If I remember correctly, there were only one or two entries in race challenge on old forums. May need separate divisions "flying" and "no jets, no propellers, no mass drives".

ACL as it is now still has a lot of things to try:
Swarms - we still have not seen much teamwork. Ant seems to only have friendly fire protection in targeting system. Same for Nest's missiles.
Unguided grenade weapons - Catapults. Rockets. Spin a grenade, roll it forward, wait, detonate, laser-ram the survivors.
Grabbing devices on laser bots - stick some thin parts into opponents armor. Maybe try to lift his front to attack bottom. Attach a grabber with jet and run jet at full thrust to get opponent out of the map? Maybe not the most practical thing to do but would be fun to look at.
Planes.
Non-plasma hovers - why is plasma the only weapon ever used in the air? Most bots are vulnerable to direct top-down laser attack. Automated bombing is possible in controlled environment, somebody tried it on old forum.

I think the main limit of ACL is that automated combat bots are hard to build and require much time.

z26

  • Experienced Member
  • ***
  • Posts: 108
  • Likes: +0/-0
    • View Profile
Re: Autonomous Combat League
« Reply #80 on: April 07, 2016, 08:48:26 pm »

Quote
Perhaps now would be a good time to add some new variations on the basic ACL to keep things interesting.
Sumo - there are infinite forces, infinite masses and partially broken physics. It will need rules to control this.
Air - would be nice. I am not sure if fights in a spherical arena would be easy to watch. My zero gravity big bot construction map (a closed box) is quite unpleasant place, camera is bumping into walls and has weird direction controls.

Some rules that could be used
no downforce.
instead of partcount classes weight classes.  If you use mass varying continuums you have to put a cap on their collective value and whatever cap value you chose is whats counted in the mass of your bot.
no hook attaching.

Nonetheless with these rules small hyper dense turtle bots that stalls might reign, so maybe forbidding continuum mass over 1 would be the way to go...  one things for sure unless rules are made to prevent this, all the bots will end up pancake shaped.  Making bumpy terrain so that a compromise between low center of mass and offroad worthiness must be achieved might be the key?

The problem is that the recipie of a good sumo bot is quite simple so this could get dull.  The normal acl reward more diverse tactics.

Also for the camera problems... maybe adding a 0.001 gravity would help?
« Last Edit: April 07, 2016, 10:16:55 pm by z26 »

cupid_the_conqueror

  • Global Moderator
  • Experienced Member
  • *****
  • Posts: 89
  • Likes: +0/-0
  • A filthy casual
    • View Profile
    • the truth of rawbots
Re: Autonomous Combat League
« Reply #81 on: April 07, 2016, 09:22:52 pm »
I think the main limit of ACL is that automated combat bots are hard to build and require much time.

There ya go! Hit the nail on the head with that one! the very nature of rawbots is slow and tedious. Combat bots take days to make even for  simple ones!

I'm telling ya' pre-programmed behaviors would have saved us TONNES of time (and well, I kinda already do that... just wait for my bot to come out in the next few days! I stole parts from literally every single person on this forum)
At the end of the day, love conquerors all.

z26

  • Experienced Member
  • ***
  • Posts: 108
  • Likes: +0/-0
    • View Profile
Re: Autonomous Combat League
« Reply #82 on: April 07, 2016, 09:30:44 pm »
In my opinion he reason making bots is so tedious is because of the sensors and programming is too limited, therefore needing to be quite clever to overcome the limitations.  Kinda like how programming done on old hardware use a bunch of tricks to be able to fit in the very limited memory available.


MarvinMan

  • Automation Expert
  • Experienced Member
  • ***
  • Posts: 236
  • Likes: +4/-0
    • View Profile
Re: Autonomous Combat League
« Reply #83 on: April 07, 2016, 10:06:53 pm »
Naturally, sumo bots will need a few rules to prevent unstoppable forces and immovable objects. My main concern that it could easily turn into a less interesting version of the current ACL with the bots just ramming each other head-on until one of them rolls or gets pushed out of the arena.

I had run a very basic test of a spherical combat zone under normal gravity just to see how hard it was to keep a plane in the intended region. Seeing the speeds that the test plane was operating at, it's likely that aerial combat wouldn't work as is would be nearly impossible to hit anything without equipping each bot with a massive amount of fire power.

I dug out the old race track map, and my original bot still works as well as used to. There was one other entry, but I don't know which blueprint it is. This version of ACL should be a lot more feasible now that XFM is standard, as the previous bots were speed limited to prevent the wheels from coming off. The biggest problem I can predict at the moment is bots ignoring the race part and just ramming the opposition into submission at the first opportunity.

z26

  • Experienced Member
  • ***
  • Posts: 108
  • Likes: +0/-0
    • View Profile
Re: Autonomous Combat League
« Reply #84 on: April 07, 2016, 10:16:16 pm »
the tighter the circle, the slower the bots are forced to move.

battlebots style (non-sumo) real life combat had trouble with too many pushing type bots that were effective but boring to watch.

Sleeping_Owl

  • Member
  • **
  • Posts: 18
  • Likes: +1/-0
    • View Profile
Re: Autonomous Combat League
« Reply #85 on: April 07, 2016, 11:16:33 pm »
I have blueprint for my race challenge entry. It was built with XFM installed.
https://drive.google.com/open?id=0ByzpYkai1-8laFFMSGw2UjMxZDQ

A race video I uploaded on old forums is still on youtube.

MarvinMan

  • Automation Expert
  • Experienced Member
  • ***
  • Posts: 236
  • Likes: +4/-0
    • View Profile
Re: Autonomous Combat League
« Reply #86 on: April 07, 2016, 11:36:46 pm »
I don't know whether you'd be able to make the arena small enough to keep speeds down without sacrificing the complex maneuvers that make aerial combat interesting.

Regarding the difficulty of building combat bots, I think most of the people building entries have a couple of basic platforms and turrets that they use as a template to build a particular bots specialised parts around.

I remember that racing bot. It looks like it's probably faster than the later versions of my racing bot that have extra downforce fins.

Re-starting the automated bot racing (Automated Racing League?) seems like the most interesting and viable concept at the moment. Here are the basic rules based on what the last version used:
Aerodynamic downforce only- no jets or propellers
Bots must be limited to a maximum width/length
Mass must be constant
Bots must have the same control continuum/start button setup as for ACL entries
No damage causing weapons
It may be necessary to ban intentional flipping (or ban intentional contact entirely) as it would probably be difficult for flipped bots to know what direction they are meant to be driving around the track.

cupid_the_conqueror

  • Global Moderator
  • Experienced Member
  • *****
  • Posts: 89
  • Likes: +0/-0
  • A filthy casual
    • View Profile
    • the truth of rawbots
Re: Autonomous Combat League
« Reply #87 on: April 08, 2016, 12:03:30 am »
I have a theory for aerial combat and how it could be done!

Planes would need to have a freeflying mode (where they fly in some pattern or with some form of randomness)

and they would need an attack mode (where they track and attack what they spot)

let them loose on the mid-sized stardust map (the water one? )

see which one finds and destroys the others 1st!  (or just toss 5-8 of Marvins Pidgeons  in the air and see which auto combat flight bot can kill the most)

it would be slow to watch, but sped up / cut it would be interesting! eps if they programmed the flying bots to attack / evade targets on its tail
At the end of the day, love conquerors all.

Sleeping_Owl

  • Member
  • **
  • Posts: 18
  • Likes: +1/-0
    • View Profile
Re: Autonomous Combat League
« Reply #88 on: April 08, 2016, 12:14:29 am »
I think most of the people building entries have a couple of basic platforms and turrets that they use as a template to build a particular bots specialised parts around.
It still takes much time. Probably it is worse for me, since I usually try to use new weapon and tactic I never used in ACL.

Here are the basic rules based on what the last version used:
Aerodynamic downforce only- no jets or propellers
Bots must be limited to a maximum width/length
Mass must be constant
Bots must have the same control continuum/start button setup as for ACL entries
No damage causing weapons
Old rules banned jets/propellers completely. Not only for downforce, but also steering, acceleration, whatever else. I think this ban was good and we should keep it.

Why constant mass? To prevent bots from increasing mass before ramming others? Using mass to balance bot on high speed turns didn't really help me, if I remember correctly, side continuums on my race bot were originally added for this.

Maybe we should move this discussion to another thread?

cupid_the_conqueror

  • Global Moderator
  • Experienced Member
  • *****
  • Posts: 89
  • Likes: +0/-0
  • A filthy casual
    • View Profile
    • the truth of rawbots
Re: Autonomous Combat League
« Reply #89 on: April 08, 2016, 01:58:14 am »
I present to you my 1st AWFUL combat bot! GoblinABRT

if the arena is bumpy, or if the angle of attack is weird, of if its busy trying to turn around. It'll miss no qualms about it! BUT IF IT HITS, GG MF*R

weighs in at 35 parts (I believe, I counted it by hand so I might be off a part or two)


https://dl.dropboxusercontent.com/u/38744980/bp_goblinABRT

for the next one, I think I'll modify the HailStorm chassie as its more stable, and I have more room for hooks to determine when it is and ISN'T okay to fire the rocket (like when climbing a hill!) 
« Last Edit: April 08, 2016, 02:00:39 am by cupid_the_conqueror »
At the end of the day, love conquerors all.