Poll

What do you like doing most in RA2/RA3/robot combat games?

Building robots
Local single player battles with manually-controlled robots
Local multiplayer battles  with manually-controlled robots (PvP with controllers or a shared keyboard)
AI-only tournament battles
Other?  Please comment below.

Author Topic: Robot Rumble 2.0 - Robot Combat Simulator - Under Development  (Read 187524 times)

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1380 on: January 09, 2020, 06:19:32 AM »
Okay, so... In this build, my verts work quite well, but there are issues. Note, this is with the sliders set to what feels like the best middle ground for my bots.

Awesome!  What are your slider settings?

Circumvolution seems to have too little gyro (large lightweight flywheel), I have a 30kg drum that seems to restrict its bot's mobility a tad too much, and smaller lighter drums seem too fragile.

Ugh.  High speed is the bane of the physics simulation.  The faster the things spin the more cheating (mass drop) we have to do, and the larger the discrepancy between robots.

Then of course there's Roller, which along with un-wheelfixed Panic Attack still has crippling problems even when idle.

I will take a look at Roller again.  It might be that the robot is so old that it is beyond fixing...

Front and overhead horizontals seem better than they have been, but... maybe a little too hindered for mobility.

My shell and ring do better in this build than a lot of previous ones. They feel about right at the moment, the weapons being powerful but still as vulnerable as they should be to being rushed.

Woohoo!  Physics for the win!

On the whole, impact seems pretty good for heavier weapons, but I feel like speed isn't imparting as much energy as it really should compared to mass, making the heavier spinners objectively better right now than the faster ones.

My first guess is that this is related to spinner bite.  Fast spinners suffer from bite problems, and you really have to drive fast at an opponent to get a good bite.  That being said, spinners need to feel good in the game regardless of what spinners are like in real life.

On a side note: when Wheely Big Cheese is inverted and fires its flipper, it launches itself close to three metres into the air. I'm wondering if this has to do with its wheels (which for authenticity and balance I never wheelfixed) being made lighter?

I'm not sure.  The makeup mass added to the chassis should offset the light wheels.  However, we are working within the limitations of a discrete physics engine at 200 ticks per second.  This can create weird interactions between rigidbodies.  I.e. For an inverted robot the chassis pushes the wheels up, then a short time later the wheels pull the chassis up.  If you have a chain of more than 2 rigidbodies (srimech -> chassis -> wheels) stacked on top of each other this whole process can get really wonky with large impulses.

Offline CyarSkirata

  • Antweight
  • Posts: 114
  • Rep: 0
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1381 on: January 09, 2020, 07:02:13 AM »
Damage is at 50
Pushout is 0
Impulse is 100
Minimum mass is 50
That's the best way I could get all of my non-bugged spinners to be at least be useable, since they're mostly at extremes one way or the other.

As for the speed thing, the problem for me is with Circumvolution. 16kg flywheel, does less than 2000 RPM, but it's got such a large diameter that the teeth do 152mph. The way it's built they have no problems engaging with the target bot, but they just don't seem to impart much energy despite the whole wheel stopping.
I occasionally get a good throw, but mostly it just does little bounces. And where it used to lever the robot forward as it accelerated, it just doesn't now.
Stuff I Done Competed In:

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1382 on: January 09, 2020, 07:46:46 AM »
Damage is at 50
Pushout is 0
Impulse is 100
Minimum mass is 50
That's the best way I could get all of my non-bugged spinners to be at least be useable, since they're mostly at extremes one way or the other.

As for the speed thing, the problem for me is with Circumvolution. 16kg flywheel, does less than 2000 RPM, but it's got such a large diameter that the teeth do 152mph. The way it's built they have no problems engaging with the target bot, but they just don't seem to impart much energy despite the whole wheel stopping.
I occasionally get a good throw, but mostly it just does little bounces. And where it used to lever the robot forward as it accelerated, it just doesn't now.

Have you tried 100% pushout? I'm a little uncomfortable setting it to 0%, for fear that we might have the situation where 2 spinners get locked inside each other.  Then again, that is unlikely as long as we keep the bite threshold low.  The "Good Hit" threshold is currently set to bite > 1 mm.

I haven't mentioned it before, but I have been tweaking one more number.  It represents how quickly the minimum mass is reached.  In the 2019 builds a 1000 RPM spinner would have 50% of its original mass.  In the 08January2020 build, a 200 RPM spinner would have 50% of its original mass.  I suspect the quicker mass drop is what is causing Circumvolution to no longer lever the robot forward upon sudden acceleration.

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1383 on: January 09, 2020, 09:36:44 AM »
I noticed this glitch in the 03January build:

EDIT: The robot that is gyrodancing has no spinning weapons.

By any chance is this working correctly in the 08January build?

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1384 on: January 09, 2020, 10:01:42 AM »
CyarSkirata,

I just had a look at Roller.  I think it is having problems with having so much mass being off-center from the robot.  I suspect that small rounding errors are throwing it off balance which cause the entire robot to wobble and shake.  The problem went away as I reduced the mass of the spinner below 10%.

We might be bumping up against a limitation of the physics engine.  The tradeoff seems to be between having spinners that feel nice and weighty, but are unstable, vs spinners that spin reliably without issues but don't have the gyroscopic effects.  We can always increase the "kick" when it hits something, so maybe it is a tradeoff we can live with?

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1385 on: January 09, 2020, 10:59:05 AM »
So it appears that once angular momentum gets over approximately 2 kg * m^2 / second, things get wonky.  I've confirmed this on Mental Breakdown and Roller.  Mental Breakdown tops out at around 5, and Roller tops out at around 7.  It looks like I need to enforce mass reduction to make sure these levels aren't reached.

Offline CodeSilver23

  • RR2 Enthusiast and Design Innovator
  • Heavyweight
  • Posts: 561
  • Rep: -2
  • Stars v Legends founder, Inventor of the Wheel-fix
    • https://www.youtube.com/c
    • View Profile
    • Youtube Channel
    • Awards
  • See profile for gamer tags: Yes
  • Discord: CodeSilverGaming
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1386 on: January 09, 2020, 11:26:26 AM »
My brain is melting.
Accomplishments:

Offline F1Krazy

  • Put me on a pedestal and I'll only disappoint you
  • Super Heavyweight
  • Posts: 920
  • Rep: 4
  • Tell me I'm exceptional, I promise to exploit you
    • View Profile
    • Awards
  • See profile for gamer tags: Yes
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1387 on: January 09, 2020, 11:43:27 AM »
My brain is melting.

Welcome to game development! It's 1% fun and 99% slamming your head against a concrete wall.
My DSL Showcase


Offline Team Lightning

  • Addicted to combat robotics, good at annoying people, and wanting to get to -10000 rep XD.
  • Beetleweight
  • Posts: 168
  • Rep: -4
  • Education is important, robots are importanter.
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1388 on: January 09, 2020, 12:04:19 PM »
We might be bumping up against a limitation of the physics engine.  The tradeoff seems to be between having spinners that feel nice and weighty, but are unstable, vs spinners that spin reliably without issues but don't have the gyroscopic effects.  We can always increase the "kick" when it hits something, so maybe it is a tradeoff we can live with?
I'm for unstable spinners, because they are more realistic.

Offline CyarSkirata

  • Antweight
  • Posts: 114
  • Rep: 0
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1389 on: January 09, 2020, 12:55:13 PM »
CyarSkirata,

I just had a look at Roller.  I think it is having problems with having so much mass being off-center from the robot.  I suspect that small rounding errors are throwing it off balance which cause the entire robot to wobble and shake.  The problem went away as I reduced the mass of the spinner below 10%.

We might be bumping up against a limitation of the physics engine.  The tradeoff seems to be between having spinners that feel nice and weighty, but are unstable, vs spinners that spin reliably without issues but don't have the gyroscopic effects.  We can always increase the "kick" when it hits something, so maybe it is a tradeoff we can live with?

I'm getting increasingly suspicious that Mac and PC simulate the game completely differently.
The problem I have is that even without turning on the weapons of Roller or un-fixed Panic Attack, they just can't steer well enough to so much as manoeuvre around the arena.
Stuff I Done Competed In:

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1390 on: January 09, 2020, 01:14:03 PM »
I'm getting increasingly suspicious that Mac and PC simulate the game completely differently.
The problem I have is that even without turning on the weapons of Roller or un-fixed Panic Attack, they just can't steer well enough to so much as manoeuvre around the arena.

It is possible, though I suspect that it might be more related to CPU/GPU load than PC vs Mac.

On my computer if I keep Roller's weapon off I can drive around slowly.  It skids around more than I would like, but it is okay.  It gets all bumpy within a second after I turn on the weapon.

Both versions of Panic Attack that you sent me drive perfectly fine.  A little slow, but they are big robots.

The version of Mental Breakdown that kix sent me a while ago drives like an absolutely maniac.  Super-maneuverable.  It overshoots like crazy when the computer AI is driving.

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1391 on: January 09, 2020, 01:21:31 PM »
We might be bumping up against a limitation of the physics engine.  The tradeoff seems to be between having spinners that feel nice and weighty, but are unstable, vs spinners that spin reliably without issues but don't have the gyroscopic effects.  We can always increase the "kick" when it hits something, so maybe it is a tradeoff we can live with?
I'm for unstable spinners, because they are more realistic.

The good:

I love the fact that we get crab walking and gyroscopic tilt for free as a result of the simulation.

The bad:

I hate the fact that spinners automatically unbalance themselves because of rounding errors.  Even if a spinner is perfectly balanced, if angular momentum gets too high a spinner will teleport from place to place as the restoring impulse overshoots every tick.

My challenge is that I would like to keep the good stuff but get rid of the bad stuff.  In Roller's case, the moment you turn on the spinner the robot starts hopping around.


Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1392 on: January 09, 2020, 03:27:01 PM »
kix,

I tried lowering Beta by lifting the wheels up 5 mm.  Sure enough, the bouncing stopped.  The issue is that the wheel joints have more compliance than they do in real life.  In the game, putting 50 kg or so sideways on the joint pushes the axle about 1 cm.  In real life the defection would be more like 0.001 cm.  To resolve this, the physics system overshoots slightly, causing a bounce effect.

Lowering the robot isn't a good solution, however, as it has the side effect of reducing the robots traction.  The lowered version of the robot accelerates significantly more slowly than the version you sent to me.

Unfortunately, I don't have a really good way to deal with this short of creating a "Magic Movement" system like exists in RA2 and RA3.  I really really don't want to do that.  I tried, and still have the code in place, but it completely changes the way robots drive. It complicates everything from driving to handling to weapons to component breakage.  I can try tweaking friction coefficient and the spring constant for the joints, but I don't know how far I can go without throwing off all other motion-related things in the game.


Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1393 on: January 09, 2020, 04:00:47 PM »
kix,

I spoke too soon.  I just tried increasing the Physics Velocity Solver Iteration Count from 4 to 10.  I'm seeing a small increase in CPU usage, and the jittering has mostly stopped.  I'll try to get this out to you for testing tomorrow.

Offline kix

  • RR2 dev
  • *
  • Posts: 3452
  • Rep: -3
  • H
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1394 on: January 09, 2020, 04:07:33 PM »
kix,

I spoke too soon.  I just tried increasing the Physics Velocity Solver Iteration Count from 4 to 10.  I'm seeing a small increase in CPU usage, and the jittering has mostly stopped.  I'll try to get this out to you for testing tomorrow.
Tbh im not too fussed about the increased cpu usage as my pc is pretty good, but as long as bots stop doing the bounce thing ill be more than happy

What i also noticed about hammers/axes with motors, the bot forcefully jumps up. I thought that it was the hammer (the tip of it weighs alot), but when the tip fell off the bot did the same thing. I can show footage

Offline Team Lightning

  • Addicted to combat robotics, good at annoying people, and wanting to get to -10000 rep XD.
  • Beetleweight
  • Posts: 168
  • Rep: -4
  • Education is important, robots are importanter.
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1395 on: January 09, 2020, 04:12:54 PM »
Speaking about hammers, I tried to make a sawbot and this happened:
 
Robot Rumble 2.0 1_9_2020 4_10_14 PM.png

Could you please fix it?

Offline kix

  • RR2 dev
  • *
  • Posts: 3452
  • Rep: -3
  • H
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1396 on: January 09, 2020, 04:55:11 PM »
Speaking about hammers, I tried to make a sawbot and this happened:
  [ Quoting of attachment images from other messages is not allowed ]
Could you please fix it?
Saws dont work. Idk when will the official fix happen, but atm a "fix" would be to attach the saw on a wheel

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1397 on: January 09, 2020, 07:02:35 PM »
Speaking about hammers, I tried to make a sawbot and this happened:
  [ Quoting of attachment images from other messages is not allowed ]
Could you please fix it?
Saws dont work. Idk when will the official fix happen, but atm a "fix" would be to attach the saw on a wheel

Hmm.  I hadn’t really thought about saws.  For now can you designate a small piece of the arm as “rubber” to prevent the generation of a blur cylinder on the arm?  Once this is done the saw itself can be built as normal.

Edit - Sawbots work fine.  They just follow the spinner rules, which means they don't really "saw" so much as "hit".

Edit #2 - I just noticed that telemetry isn't coming through properly for the spinner.  This is my first go at an overhead saw on an arm and I need to do a little digging to figure out why the telemetry isn't coming through correctly.  I did get the spinner up to speed and it sent one of the orange cones flying.  I haven't checked to see if it is doing damage correctly yet.

Here's Beta modified as a sawbot:

« Last Edit: January 09, 2020, 09:26:40 PM by cjbruce »

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1398 on: January 09, 2020, 09:01:19 PM »
kix,

I just posted a new 09January bleeding edge build.  I think this one fixes the jitter for Beta.  All I did was changed the Default Velocity Solver Iterations in Physics Settings from 4 to 10.  Let me know what you think!

http://www.robot-rumble.com/bleedingedgebuilds/

Offline BATTLEMASTER

  • Antweight
  • Posts: 24
  • Rep: 0
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1399 on: January 09, 2020, 11:01:48 PM »
I'm really liking the new build.  Although I think it takes too long to completely dismember a bot made of aluminum when using a heavy blade at 200% damage and 100% physics settings (not 200%)...

 
Bot Dismemberment.jpg


Can't wait until the new damage system is implemented since it would save me time building my bots with individual chassis plates!  Although it allows me to make complex designs that are just fun to smash apart  heck

I did notice that when I smash two wheels off of a robot that it starts to jitter.  I'm not sure why, if that's just a computer-control thing or what.