Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - cjbruce

Pages: 1 2 3 4 5 [6] 7 8 9 10 11 12 13 ... 49
101
Would it be possible to go back to the Feb 16 build and then work from that? It might be easier, considering the current build has so many issues. Also, if you’re planning on doing super small bleeding edge builds, maybe take Kix’s idea and get a small team of private testers to help.

Scrapping the thousands of small changes we have made since February 16th isn't something I'm comfortable with.

I think the better course of action is to very clearly define what we don't like about the current state.  I suspect that there are probably less than 10 things that need to be addressed.  Fix them.  Move on.

102
Roger.

Either physics changes or content changes, but not both.

The problem is that sometimes new content brings new physics.  This is the case with brushless motors.  It will also be the case with batteries, pneumatics, etc.

The best solution is going to probably involve lots of very small bleeding edge builds.

103
I'm really sorry about the frustration level.  I'm doing my best to get physics to a good place.  I know you guys are putting a ton of time into the game.  You guys are the heroes of this whole endeavor.

Thank you.

104
Ive tested the game pre work (and im typing this on a bus actually)

I feel like you have not listened to me at all and instead of fixing the current issues, you have just added stuff to game.


Follow up of some sorts, to say what is good and what is bad.
Well the spinup time is slower which is technically more realistic on small diameter spinners, but the larger it is, the slower it gets, or in hobo's case that is just the motor.


The brushless has no torque whatsoever on 10:0.1 gearing and opposite, its not usable for drive at all.



Flippers dont flip, it seems like the axle doesnt move at all.


When a vertical spinner hits the floor it loses all of its torque, thus making it impossible to selfright

Sorry for the frustration!

The torque numbers going into the system are real-life values now for all of the motors at 24 volts.  The AmpFlows and ME0708's numbers and performance should not have changed.  I acknowledge the fact that the F30-400 got messed up somehow.  I will look into it.

I will definitely take a look to see what is going on with brushless drive.  I'm assuming I need to test with a 110 kg robot at 10:1 ratio.  Which wheel size should I use?

I added code to increase the chance of bite occurring.  It looks like this has had the side effect of stopping spinners.

I'm not sure what has happened to flippers, though I'm kind of missing one to test with.  Would you mind sending me one that isn't working?

Vertical spinners not self-righting is a big problem.  I need to come up with a better balance of "bite ability" vs stopping the spinner.

105
Apart from those issues above, I'll talk about something that I think is really important and some own issues.
All motors spins up really slow, especially at the beginning, both brushed and brushless act like that, examples like Rainbow Circus with etek, or Aftershock with brushless I've posted before, the energy stacks in 10 joules from start, we can even see how slow a spinner is rotating.
And as a result, spinners barely have pushing power(except the crate in test cage) and all the vertical spinners lose their ability to self-right automatically, for motors having far less torque.

Here's a strange personal issue. My flipper Equator uses F30 for drive, but it has become extremely slow in this build. Other bots with A40 or F30 drive well as before(the position of old F30 still not right tho), only this one is having some problem.
Mar 04

Mar 09

Circus

Aftershock

  [ Quoting of attachment images from other messages is not allowed ]    [ Quoting of attachment images from other messages is not allowed ]

I just checked the numbers with Rainbow Circus, and its weapon's angular acceleration is about right with an MOI of 2.38 kg-m^2.  According to my footage it is taking approximately 5 seconds to get to 1500 RPM.  With an average torque in this interval of approximately 40 Nm, it should take about 10 seconds to get to this speed.

 
Screen Shot 2020-03-10 at 5.41.53 AM.png


For some reason Aftershock is taking 8 seconds to get to 3000 RPM, rather than 5 seconds.  It also isn't getting close to its max RPM of 4800, only topping out at 3542 RPM.  I'm not sure if this is realistic or if air drag is just too high.  According to https://robotwars.fandom.com/wiki/Aftershock, the real-life version was tuned to operate at 2400 RPM, reaching this speed in under two seconds.  Maybe try a gear ratio of 5.4:1?  This should get you to 2000 RPM in around 2 seconds.  This should also double the spinner's torque and hopefully help with self-righting.

 
Screen Shot 2020-03-10 at 5.56.23 AM.png


I'm not sure what is going on with the F30-400.  That is really weird.

106
The March 9th build is up!

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

[Updates in the March 9th Build]​

[Added] Replaced 3" Brushless Motor prefab with the TP Power TP100. This is a real-life brushless motor used by many heavyweights.  At 4" diameter x 4" length, the TP100 is physically larger than the previous 3" Brushless Motor.  It also uses the realistic torque and speed numbers of the actual motor.  The previous torque and speed values for the 3" Brushless Motor were both impossibly high.  The new motor will require gear reduction to be used as a drive motor.

[Added] New Drive Motors: Beetle Gear Motor: Perfect for Beetleweights. Inspired by the gear motor a large number of UK Beetleweights use. Drill Gearbox Motor: Fantastic for Featherweights & Lightweights. Inspired by the gearbox motors taken from cheap handheld Drills a lot of UK featherweights use. Scooter DC Motor: Great for Middleweights. Just something mid powered for all robots who require it! 4S Lipo Battery: The first of many new batteries to come. Ideal for Beetleweights.​

[Added] Spinner teeth to go with the flywheel bases and shell spinner bases.

​[Added] Added a tint color for tintable components.

[Added] New Components: Extra Decorations, Weapons added + amended

[Added] Added Spinner_Wobble_Controller to spinners. As a spinner takes damage, its center of mass becomes more and more offset. This causes the robot to wobble more and more, and makes driving more difficult.​

[Changed] Added Combined_Damageable_Object.cs. This script will track combined damage to a weapon so everything breaks off together.​

[Changed] Amended the Beetle Gear Motors with various gear ratios. Also amended the power of the SG Servo to be in line with a 10kg Servo. Offers better flipping power to Beetle's.

​[Changed] Changed the NPC T-64 to make the axle sit on '0' as well as the facing on a side panel automatically face at a more convenient rotation.

[Changed] Moved body and axle of F30-400 motor so that it was aligned with the axis.

[Changed] Added code in the SpinnerCollisionHandler onCollisionStay to stop the spinner entirely if the spinner is penetrating an immovable object by more than 10% of its radius. This is necessary to prevent the spinner from floating inside the floor while still spinning. 

[Changed] Tweaked DamageableObject.cs so that parts take 3 hits to break off. Also, the minimum damage threshold for a part has been eliminated so that all damage, no matter how small, will accumulate.​​

[Changed] New mesh for the flywheel bases [Amended] Updated weights for vast majority of components to be more realistic + competitive."​

 [Bug Fix] Updated Shape_Info of Drum_Spinner components with Robot_Resources_Control_Object.​

[Bug Fix] Fixed a bug with the robot_resources_control script, that sometimes showed the wrong component when it had a problem, making fixing the right component difficult.

[Bug Fix] Fixed a problem with the default tint, where the cloned components got their tint reset to the default one. Fixed the weight of scalable/materiable components where changes in scale/material didn't count to the weight calculation.​

​[Reverted] Reverted Robot_Reconstruction.massScaleFactor back to 0.1. The increased mass of a massScaleFactor of 1.0 caused the following bugs: 1. The increased chassis weight caused the chassis to scrape on the ground. This problem can be seen in Nuclear Crisis. 2. The increased mass caused impacts to be excessive. This can be compensated in spinner impulse response. 3. The interaction between the increased mass of the chassis and the reduced mass of vertical spinners caused vertical spinners to flip backward too easily (Circumvolution and S3).

[Bug Fix] Reduced force and torque values of motors and linear actuators by massScaleFactor.

[Bug Fix] Adjusted Combined_Damageable_Object so that when a spinner breaks off it only sets the first part's damage to 0. This means the part breaks off together rather than shattering into all of its component pieces.

[Bug Fix] Fixed the problems with the axle of motors: -Fixed axles being offset from their normal positions, the axle position was for some reason not calculated correctly on motors that didn't have (1,1,1) local scale. I now set it as (1,1,1) before getting the axle initial position and then reset it after. -Fixed the axles of the new A40-300 moving to the center of the robot, in the hingejoint the "auto configure connected anchor" option was unchecked and that meant it didn't reset after the axle got it's parent changed. -Fixed the ampflow grarbox not keeping the components aligned with the axle when moving the component.​

[Bug Fix] Amended components. A40-300 is now scaled properly but shouldn't affect previous builds as the motor was just scaled.​

[Bug Fix] Fixed the MOI and KE displayed in Telemetry_Display_Controller. The numbers are now accurate. I also verified spinup time for weapons, and this appears to be accurate as well.​

107
We are replacing the 3" Brushless Motor with the TP100.  The TP100 is about 4" diameter x 4" length.

Once it is geared down it should have higher top-end power than an AmpFlow A40-300 or an NPC T64, in about half the size and weight.  It won't be usable for drive unless you gear it down first.

108
Also, please note that all of the numbers in the game are currently for 24 Volts.  We are planning to support all the way up to 48 Volts.  All of the motors will be a lot more peppy once we increase the voltage.

You will have control over voltage by controlling the LiPo "S" rating of the batteries.  We haven't figured out the UI for this in the Robot Workshop yet.

109
The goal of all of this is that you guys won't have to do a motor swap from your February 16th build, but the motor itself will be larger.
This can be a solution. Btw when will the bot scaling tool be out? Cuz we may need to upscale our bots to fit the bigger mesh of brushless.

min440303,

I'm going to use Lotus's spinner in conjunction with a TP Power TP100 brushless motor (the same motors used in a lot of heavyweights) to calibrate the new MOI and KE computations.

There is a great online calculator that shows the basic computation at http://runamok.tech/RunAmok/spincalc.html.  Here's what I'm shooting for:

Screen Shot 2020-03-09 at 11.39.45 AM.png

110
Would there be an AI pack released for these Battlebots (in RR2)?

My intent is to have AI programming be part of the in-game experience.  You can go with the default AI, or modify the code using the in-game editor.

Please note that once we get the physics, damage, and weapon systems worked out over the next few months, I intend to go back and rewrite AI with all of the new inputs.

I expect that lots of people will be posting AI code here on GTM that people will be able to copy-paste into the game for their robots.
But, does that mean that all these bots from 'Battlebots' I can find and if so, what codes do they need?

I'm not sure what you are asking.

1. We do not have permission to include any robots from Battlebots (TM) in the game.  If you want to make or download a replica of one, that is up to you.

2. The default AI code is going to be rewritten from scratch at some point in the future.  The code that currently comes with the game will be useless after the rewrite.  You won't need to download anything except for the game itself.  The code for each robot is stored in that robot's .RR2Bot file, so if you share a robot you are also sharing its AI code.

111
Tbh id first fix the fundimental issues that we are having rn before you add more stuff into the game, id say as of now, keep the feb 16 brushless, and fix the weapon phasing issue, and we would have a solid build that would be great base for tournaments and other stuff

Here's the plan:

1. We are going to create a new, larger mesh for the existing brushless motor.  The size of the mesh will be dictated by the physics necessary to get as close as possible to the torque and speed numbers we had in the February 16th version of the game.  The old mesh had a can size of 3" diameter by 1.2" long.  The new mesh will be substantially larger than 3"x1.2", but smaller than the biggest brushed motors in the game.

2. We will rename the motor to represent its new physical dimensions.  It won't be called the 3" brushless motor, but maybe the "Heavyweight Brushless Motor", or perhaps something more specific.  Ideally we will be able to use a model from a real-life motor manufacturer that fits the specs we had before.

The goal of all of this is that you guys won't have to do a motor swap from your February 16th build, but the motor itself will be larger.

PS - The next build should fix the phasing issue.

112
Would there be an AI pack released for these Battlebots (in RR2)?

My intent is to have AI programming be part of the in-game experience.  You can go with the default AI, or modify the code using the in-game editor.

Please note that once we get the physics, damage, and weapon systems worked out over the next few months, I intend to go back and rewrite AI with all of the new inputs.

I expect that lots of people will be posting AI code here on GTM that people will be able to copy-paste into the game for their robots.

113
The 3” brushless will be in the same ballpark as an AmpFlow A28-150.  Not useless by any stretch, and completely appropriate for drive once they are geared down.  It would be an absolutely killer motor for a featherweight.  It has the highest power to weight ratio of any motor in the game.
We need a six inch brushless.

I agree that we need some bigger brushless motors for heavyweight weapons. Let me talk to Wham to see what we can come up with.
And there goes the compactness of a brushless

Brushless are still more compact.  The brush assemblies add quite a bit of length to the brushed motors.

I just want to avoid something silly like adding uber-motors with physically impossible power densities.  Here's a good discussion of the brushless use in heavyweight robot combat.  If you read down a little bit, the author talks about how and why you can get higher power densities.  Whether you used brushed or brushless you are going to have to add gearing (and the associated extra space) for drive motors:

https://www.arrow.com/en/research-and-events/articles/powering-a-battlebot

114
The 3” brushless will be in the same ballpark as an AmpFlow A28-150.  Not useless by any stretch, and completely appropriate for drive once they are geared down.  It would be an absolutely killer motor for a featherweight.  It has the highest power to weight ratio of any motor in the game.
We need a six inch brushless.

I agree that we need some bigger brushless motors for heavyweight weapons. Let me talk to Wham to see what we can come up with.

115
The 3” brushless will be in the same ballpark as an AmpFlow A28-150.  Not useless by any stretch, and completely appropriate for drive once they are geared down.  It would be an absolutely killer motor for a featherweight.  It has the highest power to weight ratio of any motor in the game.

116
Here's the file of the million joule lightweight:
  [ Quoting of attachment images from other messages is not allowed ]

Thanks for this!

With the proper numbers for the 3" brushless motor the robot becomes a lot less usable.  2/3 of its weight is in the weapon, and the weapon spins up really slowly, taking about 10 seconds to get to 3000 RPM.  It maxes out at 300,000 Joules, but in combat it is typically hitting with about 100,000 Joules because of the long spin-up time.

I need to do some hand-calculations to estimate the moment of inertia, but I suspect it is a little off.

EDIT - I did the calculations.  The MOI is off by a factor of 10x or so.  Spinup time is much too short in the game. Correcting the MOI will drop the energy way down.  This 0.5 meter diameter x 13 mm thick steel spinner should have an MOI of approximately 0.7 kg * m^2 and a peak KE of approximately 60,000 Joules.  It should only be at a few hundred joules and a few hundred RPM after 3 seconds, taking about 150 seconds to get to top speed.  It should be completely useless as a weapon in robot combat.

117
I just went back and verified the torque and speed numbers for the 3" brushless and it was WAY off.  It had about 8x the amount of torque as it should have, and ran at 2x the speed.  Even though it has a 3" diameter, the motor itself is really short.  The rotors and stators are approximately 1/2 the size of the rotors and stators for an AmpFlow A28-150, so it has about 1/2 the torque.  From personal experience, the A28-150 struggles to move a 60 lb robot around the arena if it isn't geared down.  I don't imagine an ungeared motor of this size would move a 110kg heavyweight at all.

If you were planning to use a 3" brushless for drive, it will need to be significantly geared down.  Probably with about a 10:1 gear ratio.

min440303, Nuclear Crisis will need to be geared down in order to be drivable.  It doesn't move at all in its current ungeared state.

We will probably need to add some bigger brushless motors to the game.

EDIT: The next build will have the updated torque and speed numbers for the 3" brushless motor.  This is going to break a lot of robots!

118


3" brushless is good, with LW, HW and HW are all being extremely more powerful, and at least in Feb 16 brushless is doing well in all weight class. So it should have nothing to do with the brushless motor.

Actually almost all spinners have a chance to phase themselves into the floor, maybe it's just because the spinner is too powerful and push itself into the floor by the reacting force while hitting the crate.
  [ Quoting of attachment images from other messages is not allowed ]

Oof!  I haven't verified the numbers on the 3" brushless.  I'm a little afraid what I might find when I look. :)

In order to solve the phasing problem I'm pretty sure I'm going to have to reenable collisions on the spinner part colliders.  One day I hope to make a proper single compound collider out of all of them.  The tradeoff is that it will increase the physics load having all of those spinning colliders.

119
Lots of great feedback here.  You guys found bugs WAY faster than I would have.  Thank you!

For this build I disabled collisions with the all of the colliders that made up the part and am relying entirely on the (now invisible, but still there) blur cylinder and my custom-written collision handler.  This is the cause of spinners being inside of other things weirdly.  I had hoped I had solved all of the edge cases to this, but apparently I haven't.  Thank you for finding them! :)

Aluminum density = 2760 kg / m^3
Steel density =  7861 kg / m^3

Steel is about 3 x the weight for the same thickness.  I'm pretty sure Wham intends to go back and tweak things.

In previous builds, the F30-400's origin was set on the side of the motor.  This needed to be reset to the center of the motor.  This does shift the motor down. 

min440303, please let me know if this isn't fixable by shifting the motor back up and the wheels back down.

I will take a look at energies again.  I did a LOT of fiddling with physics in the past few days.  Eventually I reset the massScaleFactor back to 0.1, but this involved adjusting a lot of other things as well that were dependent upon it.  Also, the brushless motor spinner model doesn't exist yet, and I haven't checked the torque and speed numbers for the 3" brushless motor to verify that they are realistic.

Hobo Droo, I'm having trouble with the video.  Would you mind sending the .RR2Bot file with the million Joule spinner?

CodeSilver, would you mind sending me the lightweight that launches the box through the ceiling and phases itself into the floor?


120
I just posted a 04March2020 Bleeding Edge Build:

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

In this build, spinners break off together.  Damage has been severely reduced.  Spinners now wobble.  A ton of bugs were fixed:


[Added] Added Combined_Damageable_Object.cs. This script will track combined damage to a weapon so everything breaks off together.

[Added] Added a tint default color field to the component_Info script, this is for tintable components and allows to have components that aren't white by deault.

[Added] New Components: Extra Decorations, Weapons added + amended, and I also changed the NPC T-64 to make the axle sit on '0' aswell as the facing on a side panel automatically face at a more convenient rotation.

[Added] Added Spinner_Wobble_Controller to spinners. As a spinner takes damage, its center of mass becomes more and more offset. This causes the robot to wobble more and more, and makes driving more difficult.

[Changed] Moved body and axle of F30-400 motor so that it was aligned with the axis.

[Changed] Amended components. A40-300 is now scaled properly but shouldn't affect previous builds as the motor was just scaled.

[Changed] Tweaked DamageableObject.cs so that parts take 3 hits to break off. Also, the minimum damage threshold for a part has been eliminated so that all damage, no matter how small, will accumulate.

[Bug Fix] Fixed a problem with the default tint, where the cloned components got their tint reset to the default one.

[Bug Fix] Fixed the weight of scalable/materiable components where changes in scale/material didn't count to the weight calculation.

[Reverted] Reverted Robot_Reconstruction.massScaleFactor back to 0.1. The increased mass of a massScaleFactor of 1.0 caused the following bugs:

1. The increased chassis weight caused the chassis to scrape on the ground. This problem can be seen in Nuclear Crisis.
2. The increased mass caused impacts to be excessive. This can be compensated in spinner impulse response.
3. The interaction between the increased mass of the chassis and the reduced mass of vertical spinners caused vertical spinners to flip backward too easily (Circumvolution and S3).

[Bug Fix] Reduced force and torque values of motors and linear actuators by massScaleFactor. This means the real-life values can be saved in the Component_Info, but the values are reduced in the Rotation_Component_Script and Linear_Component_Script.

[Disabled] Disabled physics collisions between spinner parts and everything else. This should help reduce physics CPU time and prevent weird collisions.

[Added] Added code in the SpinnerCollisionHandler onCollisionStay to stop the spinner entirely if the spinner is penetrating an immovable object by more than 10% of its radius. This is necessary to prevent the spinner from floating inside the floor while still spinning.

[Bug Fix] Adjusted Combined_Damageable_Object so that when a spinner breaks off it only sets the first part's damage to 0. This means the part breaks off together rather than shattering into all of its component pieces.

[Bug Fix] Fixed the problems with the axle of motors: -Fixed axles being offset from their normal positions, the axle position was for some reason not calculated correctly

Pages: 1 2 3 4 5 [6] 7 8 9 10 11 12 13 ... 49