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 187531 times)

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1500 on: January 17, 2020, 07:30:19 AM »
Thanks!  It is an ongoing effort by mostly GTM community members.

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1501 on: January 17, 2020, 08:48:48 AM »

The buggy spinners started midway through the fight, when nobody was even in the pit. I could still drive the robot fine. We got this to happen a second time, but nothing after that. We don’t have a consistent method of triggering the glitch. P2 was the main source of the problem. At some point the bar started teleporting off of the axle and was able to clip through the floor and other robots. The same thing happened to P1 later, but not quite to the same extent.

Here's the bot file for Infrared. I'm not sure what physics stats we were using though, as this was on Kix's computer.

I think I got it!  When the last component is broken off an axle, the mass of the remaining axle is set to 0.000001 kg.  I incorrectly determined what the "last component" meant.  When the colored bars were broken off of Infrared, the game thought that they were the last component, and had set the mass remaining to 0.000001 kg.  What you see in the video is the underlying steel bar that is teleporting all over the place because it has almost no inertia (mass).  I fixed it in the system, and should have a new build up today.

For Infrared, I also noticed that the colored bars are attached directly to the axle and not to the underlying steel bar.  This means that the colored bars will break off as a separate piece, kind of like a thin strip of plastic.  The manual fix is to attach the colored pieces to the underlying steel bar and mark them as "decoration" material.  I'm not sure if there is a way to automatically attach the entire strip to the underlying bar and mark them as "decoration".

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1502 on: January 17, 2020, 11:33:04 AM »
I was hoping to get an Alpha build out today. 

Right now I'm trying to track down a crashing bug that occasionally happens when Infrared and Circumvolution collide.  For some reason the 08January, 12January, and current builds crash to the Main Menu when the two robots collide in a certain way.  I haven't been able to replicate the crash in the Unity Editor, so I'm going to need to do some more digging...

Offline kix

  • RR2 dev
  • *
  • Posts: 3452
  • Rep: -3
  • H
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1503 on: January 17, 2020, 12:04:34 PM »

The buggy spinners started midway through the fight, when nobody was even in the pit. I could still drive the robot fine. We got this to happen a second time, but nothing after that. We don’t have a consistent method of triggering the glitch. P2 was the main source of the problem. At some point the bar started teleporting off of the axle and was able to clip through the floor and other robots. The same thing happened to P1 later, but not quite to the same extent.

Here's the bot file for Infrared. I'm not sure what physics stats we were using though, as this was on Kix's computer.

I think I got it!  When the last component is broken off an axle, the mass of the remaining axle is set to 0.000001 kg.  I incorrectly determined what the "last component" meant.  When the colored bars were broken off of Infrared, the game thought that they were the last component, and had set the mass remaining to 0.000001 kg.  What you see in the video is the underlying steel bar that is teleporting all over the place because it has almost no inertia (mass).  I fixed it in the system, and should have a new build up today.

For Infrared, I also noticed that the colored bars are attached directly to the axle and not to the underlying steel bar.  This means that the colored bars will break off as a separate piece, kind of like a thin strip of plastic.  The manual fix is to attach the colored pieces to the underlying steel bar and mark them as "decoration" material.  I'm not sure if there is a way to automatically attach the entire strip to the underlying bar and mark them as "decoration".
This is where grouping would work

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1504 on: January 17, 2020, 05:45:04 PM »
The 17January2020 build is out:

https://robot-rumble.itch.io/builds

[Added] Added Hotkeys and Hotkeys selection screen in the Robot Workshop.  What hotkeys would you like to see to speed up your workflow?

[Bug Fix] Fixed the incorrect calculation of the location of the center of mass for components that are scaled in the workshop.  This was causing robots like S3 to tip over immediately, as well as the robot Rainbow Circus 2 to tip over once its shell had been removed.

[Bug Fix] Fixed a problem that would allow you to delete both mirrored components if one of the two was selected, but wouldn't allow it if the other was selected instead.

[Bug Fix] Once robots are Out Of The Arena (pitted or otherwise), the robot is marked as disabled and the control signals are automatically set to zero. This should prevent the crazy situation where spinning weapons still spun in the pit, turning the arena pit into a blender that kept throwing pieces everywhere.  See kix's video from a few posts back.

[Bug Fix] I had incorrectly assumed that only one object could be attached to an axle, and all other objects must then be attached to that. This resulted in me setting the mass of the entire rigidbody to 0.000001 kg even when there were other components still remaining. I now check to make sure all components are truly gone before setting the rigidbody mass to zero. This should fix the bar spinner teleportation on Infrared we saw in kix's video.

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 #1505 on: January 18, 2020, 08:24:58 AM »
Can you fix the telemetry?

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1506 on: January 18, 2020, 11:43:25 AM »
Can you fix the telemetry?

I can try!  Can you point me to a few .RR2Bot files with the symptoms?  The more, the better!

Offline kix

  • RR2 dev
  • *
  • Posts: 3452
  • Rep: -3
  • H
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1507 on: January 18, 2020, 12:35:07 PM »
yo when are you also adding the hotkeys like:
Delete button - Delets a component, basically a hotkey to the red X
Ctrl-Z - Undo a move

Offline UberPyro

  • Lightweight
  • Posts: 205
  • Rep: 6
  • Well, here goes.
  • Awards GTMCS Division Winner
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1508 on: January 18, 2020, 03:40:32 PM »
I haven't looked at this in a long time but something that allows for very precise movement with wasdqz or wasd-shift-space in amounts of ~0.01 a la movepixel was one of the things I remember wanting.

Offline TheRoboteer

  • Mr Big Blue Flipper
  • *
  • Posts: 1587
  • Rep: 9
  • "I HATE SEBASTIAN COE!"
  • Awards Donated money for site hosting 2021
    • View Profile
    • Awards
  • Discord: TheRoboteer#8784
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1509 on: January 18, 2020, 03:44:56 PM »
I haven't looked at this in a long time but something that allows for very precise movement with wasdqz or wasd-shift-space in amounts of ~0.01 a la movepixel was one of the things I remember wanting.
You can edit the coordinates of components to like 3 decimal places
Notable Tournament Performances:
The Beginner's Guide to IRL:
https://gametechmods.com/forums/tutorials-and-tips/irl-a-beginner's-guide

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 #1510 on: January 18, 2020, 04:07:53 PM »
Can you fix the telemetry?

I can try!  Can you point me to a few .RR2Bot files with the symptoms?  The more, the better!
I mean the glitch where the telemetry screen doesn't turn off when the telemetry button is pressed.

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1511 on: January 18, 2020, 04:39:04 PM »
Just solved a lingering bug with spinners that caused the game to crash back to the main menu during a fight.  I don’t know how much of a problem it was for you guys but it was driving me nuts not being able to fight two spinners against each other.

I think that is the last of the crashing bugs.

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1512 on: January 18, 2020, 08:26:40 PM »

And while I'm at it, here's the version of it where I was testing a crusher. You'll need to set a second one of the gearboxes to 10/0.1 to get the result where it barely steers in battle. Just like with Panic Attack, the issue doesn't effect it at all in the workshop test area.
  [ Quoting of attachment images from other messages is not allowed ]

I have been messing around with Circumvolution C, and I can't seem to get it to exhibit the behavior you were referring to.  With the gearbox set to 10:0.1, I see the same steering performance as with the gearbox set to 1:1.  Would you mind sending a copy of the robot where you see the effect?

One thing to note is that, in general, CPU usage is 4x higher in combat than it is in the test arena.  In the test arena, the computer only has to run a single robot's worth of physics.  In combat, the CPU has to run 2 x physics + 2 x AI code.  Not only that, but there are WAY more potential colliding colliders in combat.  Maybe this is the slowdown you are seeing?  Do you see a driving performance improvement if you set the physics tick rate to 100 ticks/second?

Offline UberPyro

  • Lightweight
  • Posts: 205
  • Rep: 6
  • Well, here goes.
  • Awards GTMCS Division Winner
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1513 on: January 18, 2020, 09:06:11 PM »
I haven't looked at this in a long time but something that allows for very precise movement with wasdqz or wasd-shift-space in amounts of ~0.01 a la movepixel was one of the things I remember wanting.
You can edit the coordinates of components to like 3 decimal places
I'm well aware of this, it's just that in very precise builds like my Nightmare build it's too difficult to math out where everything should be, and the build time would go down rapidly if there were movepixel-esque controls rather than having to edit the coordinate, see if its close enough, edit it again, etc.

I'm going to quote the old post I made in here, because it outlines some stuff that really ought to have hotkeys.

Is there anyway I can see what features are planned for future versions? Anyway, I have some suggestions after building in the botlab for some time. A lot of suggestions and a lot of them are complex, so I don't expect you to agree with all of them or try to implement much of what's here in the near future, I know you guys are busy.
I know you guys are not nearly done with the botlab and that given time you guys will make it great, but talking about the botlab right now as it is, it's still pretty clunky in the sense that it takes a very long time to use the controls and the tools given to you to make the botlab do what you want it to do. First of all, I think keyboard shortcuts would improve the building experience 1000%. Things like pressing x to focus the x coordinate, delete for deleting components, arrow keys/wasd + shift + space for moving things around/rotating, basically all the buttons and fields in the interface should have some keyboard mapping because clicking on everything takes forever. Remapping keys would also be great, i.e. i much prefer RMB to MMB for panning. As it is, it's very easy to make mistakes by putting the wrong numbers into the coordinates or into the wrong coordinates, and it's also way too easy to get yourself into a situation where you move something too far away and can only fix it through editing the bot file. An undue button would be a great help for this, and additionally some way to control the max distance from origin of all the components and toggle the room that you are in would all be huge helps. It's also really easy to get objects stuck inside of others, and there's currently a glitch where some small objects won't be able to be selected unless you exit and join the workshop. A lot of these issues could also be helped by having an expandable tree view of all the components like in the bot file and being able to select components like that. Being able to turn off blur from being too close would also be great. When using the moving sliders, they go in increments of .025, and if the component isn't currently at a multiple of .025 then it snaps to a multiple: there should be some way to turn the snapping to a multiple off, because a lot of the more complex builds will not use those multiples very often. This also applies to the rotation and size sliders. There should be a way to control what interval the moving axes move by, either through a field or through several hotkeys which are each set to common values, i.e. .01 and .005 with .025 as the default. It would be great if there was some way to change the directions of the coordinate system so that it doesn't have to be determined by what the object is sitting on, a lot of the times I just want the normal xyz directions. It would be fantastic if there was a way to copy the x, y, or z coordinate of one object and then set the coordinate to another in a streamlined fashion, e.g. hold x while clicking on another object. It would also be great if there was some way to clone mirrors of objects and flip each of the components. It would be nice if we could smooth components along a certain rotational axis. For the chassis and custom object builder it would be great if we could change the numerical values of the xy coordinates via text field of each of the vertices like we can for objects in the rest of the builder.

Offline CyarSkirata

  • Antweight
  • Posts: 114
  • Rep: 0
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1514 on: January 18, 2020, 09:13:10 PM »

And while I'm at it, here's the version of it where I was testing a crusher. You'll need to set a second one of the gearboxes to 10/0.1 to get the result where it barely steers in battle. Just like with Panic Attack, the issue doesn't effect it at all in the workshop test area.
  [ Quoting of attachment images from other messages is not allowed ]

I have been messing around with Circumvolution C, and I can't seem to get it to exhibit the behavior you were referring to.  With the gearbox set to 10:0.1, I see the same steering performance as with the gearbox set to 1:1.  Would you mind sending a copy of the robot where you see the effect?

One thing to note is that, in general, CPU usage is 4x higher in combat than it is in the test arena.  In the test arena, the computer only has to run a single robot's worth of physics.  In combat, the CPU has to run 2 x physics + 2 x AI code.  Not only that, but there are WAY more potential colliding colliders in combat.  Maybe this is the slowdown you are seeing?  Do you see a driving performance improvement if you set the physics tick rate to 100 ticks/second?

You need more than one of the gearboxes geared down, producing a total gear-down greater than 100/1. Try setting all three gearboxes at 10:1. So it should be 1000:1 total.

Also, while I'm here, I've still never seen the game stop considering something a spinner because it has rubber on it, but I feel like that's supposed to be a thing now.
« Last Edit: January 18, 2020, 09:39:17 PM by CyarSkirata »
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 #1515 on: January 19, 2020, 07:13:04 AM »


I have been messing around with Circumvolution C, and I can't seem to get it to exhibit the behavior you were referring to.  With the gearbox set to 10:0.1, I see the same steering performance as with the gearbox set to 1:1.  Would you mind sending a copy of the robot where you see the effect?

One thing to note is that, in general, CPU usage is 4x higher in combat than it is in the test arena.  In the test arena, the computer only has to run a single robot's worth of physics.  In combat, the CPU has to run 2 x physics + 2 x AI code.  Not only that, but there are WAY more potential colliding colliders in combat.  Maybe this is the slowdown you are seeing?  Do you see a driving performance improvement if you set the physics tick rate to 100 ticks/second?

You need more than one of the gearboxes geared down, producing a total gear-down greater than 100/1. Try setting all three gearboxes at 10:1. So it should be 1000:1 total.

Shoot.  That means I need to figure out where the rest of the gearboxes are. :)

EDIT:  Here's what I am seeing with the first gearbox at 10:1 and the second at 10:0.1



Also, while I'm here, I've still never seen the game stop considering something a spinner because it has rubber on it, but I feel like that's supposed to be a thing now.

Uh oh.  I hadn't considered that people would use rubber to circumvent weapon spinners entirely.  This is bad because it completely sidesteps the weapon damage system.  By placing rubber on a spinner, you are telling the game that this isn't a spinner -- it is a wheel.  Wheels don't get the (now invisible) Weapon_Blur_Cylinder, and so therefore can't damage anything.

I have an idea for a no-user-input-required fix that just finds the outermost component.  If the outermost component is rubber, it treats it as a wheel.  If not, then it treats it as a weapon.  This can be recomputed every time you knock off a piece, so that a spinning object can first be a wheel, then when its rubber gets ripped off it becomes a weapon.
« Last Edit: January 19, 2020, 07:38:06 AM by cjbruce »

Offline min440303

  • Lightweight
  • Posts: 192
  • Rep: 2
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1516 on: January 19, 2020, 07:46:50 AM »
Something bad happened.
When I was building something for about 2 hours, I accidentally clicked the Option 1(convex) collision generation when editing the chassis. Then my bot died.
 
unknown.png

Is there any way to bring it back to life?
 
Mad but not mad.

Stuff I kinda done at

Offline cjbruce

  • Super Heavyweight
  • Posts: 963
  • Rep: 11
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1517 on: January 19, 2020, 10:26:45 AM »
Something bad happened.
When I was building something for about 2 hours, I accidentally clicked the Option 1(convex) collision generation when editing the chassis.
Is there any way to bring it back to life?
 

Oh no!  Looking into it!

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 #1518 on: January 19, 2020, 05:54:43 PM »
I was able to replicate the Beetle Crab glitch with a beetleweight robot I made.

Offline CyarSkirata

  • Antweight
  • Posts: 114
  • Rep: 0
    • View Profile
    • Awards
Re: Robot Rumble 2.0 - Robot Combat Simulator - Under Development
« Reply #1519 on: January 19, 2020, 06:48:24 PM »
Shoot.  That means I need to figure out where the rest of the gearboxes are. :)

EDIT:  Here's what I am seeing with the first gearbox at 10:1 and the second at 10:0.1

There are three gearboxes between the e-tek and the crusher arm. One straight and two 90 degree.


Uh oh.  I hadn't considered that people would use rubber to circumvent weapon spinners entirely.  This is bad because it completely sidesteps the weapon damage system.  By placing rubber on a spinner, you are telling the game that this isn't a spinner -- it is a wheel.  Wheels don't get the (now invisible) Weapon_Blur_Cylinder, and so therefore can't damage anything.

I have an idea for a no-user-input-required fix that just finds the outermost component.  If the outermost component is rubber, it treats it as a wheel.  If not, then it treats it as a weapon.  This can be recomputed every time you knock off a piece, so that a spinning object can first be a wheel, then when its rubber gets ripped off it becomes a weapon.

I mentioned it because it doesn't happen, at all, but someone said something at some point that I took to mean that it was supposed to be a thing. So I thought it was a bug that it wasn't happening.

It would be useful to be able to make something not be a spinner though, without the extra few kg of wheelfixing. Custom wheels beyond a certain size are crippled by using the spinner physics.
Yesterday, in Parsec Rumble 2, Wheely Big Cheese danced into the pit because as soon as it left the floor it started doing weird soft gyro in the air, and when it landed it did weird moon-gravity dancing that made it hard to get both wheels back on the floor.

And while I'm on the subject of wheels: Rubber parts beyond a few centimeters on wheelfixed components (such as custom wheels) gets insta-destroyed by so much as lightly grazing another robot. Meaning - say - that it doesn't much matter how tough a steel cylinder custom wheel is, because the entire rubber tread on that wheel disintegrates once the two bots meet.
Stuff I Done Competed In: