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 ... 49
81
« on: March 12, 2020, 10:21:38 AM »
How low are you guys setting the damage slider? CodeSilver says 50%.
That will give me a number to work from.
82
« on: March 12, 2020, 10:00:43 AM »
Ok so wheelfix kind of has a purpose again, as It is able to nerf a lot of the damage.
Hmmm... Wheelfixing shouldn't be a thing anymore. How does it work?
Need to fix this...
Please don’t “fix” this just yet. At least not until the damage is fixed.
I can't get rid of it if I don't know what it is. It sounds like an exploitable bug though. I'm in bug squashing mode! The right answer is to address the damage problem, then get rid of the exploit.
83
« on: March 12, 2020, 09:59:20 AM »
Ok so the weapon wobble works... Not too great. I feel like wobble shoudnt appear if youre bot is the one that is dealing the damage to the opponent. It should only appear if you are hitting the floor at high speed or if your spinner is receiving high ammount of damage on the disc itself. Also This bot tends to lose the weapon while hitting the floor at high speed.
I think you might be right. If a spinner hits an opponent, it has presumably been engineered to take the shock load. I will try reducing/eliminating the damage that occurs to self. EDIT - I reduced selfDamageTaken to zero when you hit an opponent. You should see the changes on the next build. You can still take damage from hitting things that aren't other robots.
84
« on: March 12, 2020, 09:57:44 AM »
Ok so wheelfix kind of has a purpose again, as It is able to nerf a lot of the damage.
Hmmm... Wheelfixing shouldn't be a thing anymore. How does it work? Need to fix this...
85
« on: March 12, 2020, 09:33:26 AM »
min440303,
Thanks for the extensive writeup! Lots of great video to unpack here.
I think I need to look in two main places:
1. The spinner collision handler code. I need to make spinner hits more likely to register as a "good hit". Right now it is too likely that a "glancing blow" will be registered when a spinner hits an opponent. Glancing blows don't do a whole lot. They should throw sparks, but even that isn't happening much.
2. The spinner weapon breakage code. I need to consolidate damage so that all damage is taken by the entire spinner. Right now it is still possible to break off a piece of the spinner and have it get lodged inside the robot.
86
« on: March 12, 2020, 09:03:08 AM »
Just put up the 12March2020 bleeding edge builds: http://www.robot-rumble.com/bleedingedgebuilds/[Added] Added more conditions for sparks to occur when the spinner hits the floor. It seems to be tuned to occur very rarely though. [Changed] Reduced the amount of energy lost when a spinner hits an immovable object to 1/10th its value that occured in the previous build. It is starting to feel a lot better now. [Changed] Reduced stall torque and no-load RPM for the Motenergy ME0708 so that it is correctly rated for 24 volts. Added a "Max Safe Torque" of 34 Nm. This corresponds to roughly 400 Amps, which the motor can safely pull for 1 minute. [Changed] Reduced stall torque and max RPM of Pancake Etek motor to be consistent with the real-life LEM170-D127 motor. Added a max safe torque of 50 Nm.
87
« on: March 11, 2020, 09:26:54 PM »
It is simple to set individual tournament limits. I was referring to the limits on components. We don’t currently have anything that goes over 48 volts in the game.
88
« on: March 11, 2020, 09:25:03 AM »
I have extrapolated the torque-speed curves out at 48 volts. It looks like we are going to run into a few interesting effects/issues:
3. At 48 volts, the brushless motor is current-limited for most of its operating range. This means it will provide the same 8.5 Nm of torque all the way up to 20,000 RPM. A solenoid-operated ME0708 will provide 70 Nm of torque at the start, but will really rapidly drop off.
Whats an IRL example of 8.5 Nm?
8.5 Newton-meters is equivalent to the following: 1200 ounce-inches 6.7 foot-pounds 75 inch-pounds At a distance of 1 meter, the motor could apply a force of 8.5 Newtons (approximately 1.9 pounds). At 10 cm, the motor could apply a force of 85 newtons (approximately 19 pounds). In real life for most applications, you would gear the motor down to increase the torque and decrease the top speed. If you wanted to be able to lift a 110 kg heavyweight completely off the ground with a 70 cm long lifter arm, you would need: 110 kg * 9.8 m/s^2 * 0.70 meters = 755 Nm of torque To do this with a brushless motor capable of 8.5 Nm of torque, you would need a total gear ratio of at least 90 : 1. In real life, this probably would mean 6-7 stages of reduction. You would probably want to double this gear ratio just to give yourself some wiggle room. The A28-150G speed reducer uses 3 reduction stages to achieve an 8.3 : 1 gear ratio. We just hid all three chains inside the box so you can't see them. :) EDIT: Changed the calculation to account for the fact that heavyweight robots are not point masses. You would have to lift their center of mass, which is going to be about 70 cm from the axis of your motor.
89
« on: March 11, 2020, 07:53:15 AM »
I have extrapolated the torque-speed curves out at 48 volts. It looks like we are going to run into a few interesting effects/issues: 1. The AmpFlow A40-300 is only rated to 24 volts. Unless we can figure out how to do separate batteries for drive (24 volts) and weapons (48 volts), the A40-300 won't be an option for robots wanting a higher voltage for their weapons. We don't have any brushed drive motors in the game that can operate at 48 volts. I believe a lot of the heavyweight robot builders are running into this problem as well. 2. The ME0708 (or any of the really big motors) will exceed the current capacity of the available ESCs at 48 volts. This means we will need to put on-off solenoids into the game for the really big motors. They will only be controllable by an on-off toggle switch at 48 volts, and won't be reversible. 3. At 48 volts, the brushless motor is current-limited for most of its operating range. This means it will provide the same 8.5 Nm of torque all the way up to 20,000 RPM. A solenoid-operated ME0708 will provide 70 Nm of torque at the start, but will really rapidly drop off.
90
« on: March 11, 2020, 06:46:24 AM »
PS - I'm still working out the numbers on some of the newer motors. In real life, Aftershock uses an LEM170. It is a 7" diameter motor slightly smaller than an Etek, but quite a bit larger than the 4" brushless in the Aftershock rep. For now, the "Pancake Etek" is the closest thing in game to an LEM170.
You mean LEM170 is the same size as Pancake Etek? That means my Aftershock rep is incredibly small to fit this motor...
It is funny that you say that because I believe that Team Shock had the exact same problem. I believe they intended to use a brushless weapon initially, but then went to the brushed LEM170 when the brushless ESC kept burning up. They had to redesign to fit the larger motor.
91
« on: March 11, 2020, 06:02:42 AM »
The scaling stuff is awesome! It will save tons of efforts to do some resizing stuff. Really appreciate it! Also, some issues that still might need to be done:
Excellent! @tashic is a genius. :) -Flippers don't work
Caveat Emptor! I have done precisely zero tuning of the pneumatic flippers. This is because we don't have a pneumatic system yet. The amount of force and speed created by the existing flippers is purely fictional. :) -Old F30s' moving downward
-Looks like spinner wobbling makes the bot drive much slower, when Aftershock's spinner took certain damage, the bot just couldn't move even when the weapon motor was off.
This is weird. The COM offset should not have affect the position of components, except as far as the joint is being stressed when the spinner is spinning. -Motor's starting torque are still really bad, thus lifters, crushers, saws and hammers are useless. And spinners spins up extremely slow at the start no matter how it's geared.
Caveat Emptor! We don't have an electrical distribution system model yet. All motors are set to 24 volts, except for the beetleweight-sized motors. Robogames and BattleBots both have a 48 volt limit. Robot Wars has a 75 volt limit. I'm not sure what our limit is going to be. Assuming we set it the in-game limit to 48 volts, any motor that can handle voltage higher than 24 volts will suddenly have twice the top speed and significantly more torque. Torque is limited by the amount of current running through the motor windings. You need to gear the brushless motor WAY down to increase its torque. This is worthwhile, however, as it has a much higher top speed than a similar brushed motor. Here's what I have so far at 24 volts (NOTE - these numbers are subject to change!): The brushless is incredibly powerful overall (torque x speed at a given RPM), but you have to gear it way down to take advantage of the power. Expect to gear it down three times as much as an Ampflow. Caveat Emptor! - We haven't started on hammers and axes yet. They are probably going to require pneumatics. PS - I'm still working out the numbers on some of the newer motors. In real life, Aftershock uses an LEM170. It is a 7" diameter motor slightly smaller than an Etek, but quite a bit larger than the 4" brushless in the Aftershock rep. For now, the "Pancake Etek" is the closest thing in game to an LEM170. -Spinners easily lose all its speed before delivering a hit.
I'll work on this some more today. There needs to be some speed loss, but I think it is still too much. When I flip aftershock, I'm see KE losses of between 1%-25% every time it hits the ground. In real life I think the numbers should be somewhere between 0.1% - 2%. -Equator's driving issue with F30s
I am at a loss on this one. I will need to do some more poking around to see what is happening. I suspect it might be because the motor is dragging on the ground, but I'm not sure.
92
« on: March 10, 2020, 09:21:04 PM »
Re-uploading again. This time I changed the filenames slightly:
RR2-Windows-10March2020-BleedingEdgeTestBuild2
My internet connection is a little slow on uploads. It might take a few minutes before they appear on your end.
PS - kix, I just verified that the Flipper Piston System isn't working. I have no idea why. The script is getting the control signal, and the torque and speed numbers are insanely high. I will have to do some poking around tomorrow.
EDIT: The re-upload is complete. They should be ready for download.
93
« on: March 10, 2020, 08:57:28 PM »
We cannot unzip the file
Sorry about that! Please try re-downloading. Filezilla only got part of the .zip file uploaded before the server filled up. I believe that both the MacOS and Windows .zip files were corrupted as a result. EDIT: The MacOS .zip file should be 313 MB. The Windows .zip file should be 311 MB.
94
« on: March 10, 2020, 08:38:12 PM »
Re-uploading. Apparently our server was full. I had to delete some old builds...
95
« on: March 10, 2020, 08:28:04 PM »
The 10March2020 Bleeding Edge Builds are up: http://www.robot-rumble.com/bleedingedgebuilds/Bonus! @tashic got the scaling function working! Scaling is only possible from the "Electronics" tab in the workshop. To scale, use the hotkey "Shift-E". It only scales the scaleable components, like shapes. Motors, electronics, etc. are not scaled. If you scale down, you could end up with components inside other components.
96
« on: March 10, 2020, 04:37:25 PM »
Not yet. It’s on the list though. Maybe tomorrow if you like what you see in this one?
97
« on: March 10, 2020, 03:46:09 PM »
Building with a few changes now. I'm going to try to get a bleeding edge build out tonight with the following changes:
[Changed] Increased default torque and RPM numbers for the A28-150 to match the values for 24 volts. They were previously set for 12 volts.
[Changed] Reduced spinner wobble by setting the max COM offset to 10% of radius. Previously it was set to 20% of radius.
[Changed] Less spinner speed is lost on impact. Instead of a "good hit" causing all KE to be lost, only a fraction of it is lost. This fraction depends on how far the spinner overlaps the floor or other robot. If it overlaps by half of the spinner's radius, then all energy is lost. Otherwise, only some speed is lost.
[Added] Added a "Flip Robot" button in the Robot Workshop Test Cage. This helps to simulate being launched by a flipper and for testing self-righting mechanisms.
98
« on: March 10, 2020, 02:39:11 PM »
I put in a "flip robot" button for the Robot Workshop test cage. It should help when we want to test self-righting systems.
In particular, I have been testing it out with min's Aftershock rep with the TP100 motor. I have found that Aftershock is not capable of self-righting unless the weapon is already spinning when it flips over.
Does anyone know if this is accurate to the real robot? Can Aftershock self-right from a cold start?
Aftershock is not capable of self righting from a cold start. However once it hits the floor it fully stops. Its like the disc has no mass
My hand calculations indicate that Aftershock's motor should need about 10x more torque than the TP100 is capable of producing to self-right from a cold start. This matches what I see in the game. My hand calculations also indicate that if Aftershock has enough angular momentum (about 200 kg * m^2 / s^2) it should be able to bounce once to a height of 15 cm, then bounce again and flip itself over. In the fight between Aftershock and Eruption, Aftershock used tiny little hits to get itself back upright again: The fight starts around 28:00. I think kix is correct. It isn't a matter of increasing the motor torque. Rather, the correct approach looks to be not completely killing the rotation on each hit, even on the so-called "good hits". There is too much blur in the video to measure precisely how much speed is lost, but I might be able to figure it out based on how high Aftershock hops when its spinner hits the ground at 28:05.
99
« on: March 10, 2020, 02:01:38 PM »
I put in a "flip robot" button for the Robot Workshop test cage. It should help when we want to test self-righting systems.
In particular, I have been testing it out with min's Aftershock rep with the TP100 motor. I have found that Aftershock is not capable of self-righting unless the weapon is already spinning when it flips over.
Does anyone know if this is accurate to the real robot? Can Aftershock self-right from a cold start?
100
« on: March 10, 2020, 11:52:09 AM »
No worries guys. The criticism is welcome, and I really do appreciate the feedback that everyone is giving. This is a really tight spot in the development because we have lots of new content with a physics system that needs to be tuned. Some of the time that means dialing back and breaking things that worked previously. Criticism is a super-important part of this process. You guys are on the cutting edge, and we are going to have to iterate the  out of this to get it right.
Pages: 1 2 3 4 [5] 6 7 8 9 10 11 12 ... 49
|
Show Posts