Go to Post That little voice in your head is not Ari telling you to keep going. It is your guardian angel telling you to stop! - Al Skierkiewicz [more]
Home
Go Back   Chief Delphi > CD-Media > Photos
CD-Media   CD-Spy  
portal register members calendar search Today's Posts Mark Forums Read FAQ rules

photos

papers

everything



Subjective Eval of Common Drive Systems

By: Andrew Schreiber
New: 03-20-2017 03:22 PM
Updated: 03-20-2017 03:22 PM
Views: 714 times


Subjective Eval of Common Drive Systems

A simple subjective evaluation I decided to toss together as both an example of how folks could represent design choices as well as to help students evaluate options or at least provide a jumping off point for research.

Recent Viewers

  • Guest

Discussion

view entire thread

Reply

03-20-2017 05:13 PM

josesantos


Unread Re: pic: Subjective Eval of Common Drive Systems

What is a "2+2" drive system?



03-20-2017 05:15 PM

frcguy


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by josesantos View Post
What is a "2+2" drive system?
Two omnis, two traction. 971 has one this year.



03-20-2017 05:21 PM

Nate Laverdure


Unread Re: pic: Subjective Eval of Common Drive Systems

Can we call parallel omnis "Asteroid Drive" ?



03-20-2017 06:04 PM

Dillon Carey


Unread Re: pic: Subjective Eval of Common Drive Systems

I'm not sure how your defining power. I'd argue that it's pretty independent from drive train type (except if your accounting for efficiency losses).
For example, you have 2+2 listed as low power. Our 2014 bot had 6 CIMs, which most would argue is plenty (or more) power than you need.



03-20-2017 06:21 PM

Oblarg


Unread Re: pic: Subjective Eval of Common Drive Systems

I confess, I'm not at all a fan of 2+2. The turning center is nowhere near the center of the robot, which makes maneuvering very strange. It also has a nasty tendency to fishtail. Calling it "more mobile" than a standard 6wd is...a bit of a stretch.



03-20-2017 06:28 PM

snoman


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Dillon Carey View Post
I'm not sure how your defining power. I'd argue that it's pretty independent from drive train type (except if your accounting for efficiency losses).
For example, you have 2+2 listed as low power. Our 2014 bot had 6 CIMs, which most would argue is plenty (or more) power than you need.
Not sure but I am guessing something like pushing power. All other factors equal. Maybe



03-20-2017 06:30 PM

Michael Hill


Unread Re: pic: Subjective Eval of Common Drive Systems

KoP drive gets dinged on complexity? That thing can get done in a couple of days if not within a few hours. Have any Kitbots really broken yet?



03-20-2017 06:34 PM

snoman


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Michael Hill View Post
KoP drive gets dinged on complexity?
The KOP is the baseline I think

That thing can get done in a couple of days if not within a few hours. Have any Kitbots really broken yet?
I doubt it they are extremely tough



03-20-2017 06:37 PM

Andrew_L


Unread Re: pic: Subjective Eval of Common Drive Systems

I like this - it's a nice visual guide for those who would be confused by a weighted cost/benefit table.

My only question is what examples did you use as a baseline for this? Of course it's subjective, so this is all based on a team ability/skillset/resource capability that you had in mind, so it would be interesting to have that information on the graphic for context.



03-20-2017 06:47 PM

frcguy


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Michael Hill View Post
KoP drive gets dinged on complexity? That thing can get done in a couple of days if not within a few hours. Have any Kitbots really broken yet?
I think you're reading it wrong. KoP isn't getting dinged on anything.



03-20-2017 07:17 PM

Schroedes23


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by snoman View Post
Not sure but I am guessing something like pushing power. All other factors equal. Maybe
In terms of pushing power take a look at the matches I've posted below:
https://youtu.be/i9-eH3Iym38
https://youtu.be/imPp9DMwEOg?t=42s

A 2+2 can have pushing power.



03-20-2017 07:23 PM

EricH


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Oblarg View Post
I confess, I'm not at all a fan of 2+2. The turning center is nowhere near the center of the robot, which makes maneuvering very strange. It also has a nasty tendency to fishtail. Calling it "more mobile" than a standard 6wd is...a bit of a stretch.
OTOH, if you do it right, you can use that fishtailing to maneuver around one end of the robot, all the time, which might be important if you are trying to maneuver a tall, unstable stack of 5 totes +1 can around. If your traction wheels are around the stack, it simply spins...

It can be done.

There's also the 2x2 which would be putting omnis on opposite corners paired with tractions on opposite corners.



03-20-2017 07:27 PM

EmileH


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Schroedes23 View Post
In terms of pushing power take a look at the matches I've posted below:
https://youtu.be/i9-eH3Iym38
https://youtu.be/imPp9DMwEOg?t=42s

A 2+2 can have pushing power.
+1 to this. 1625's 2014 drivebase was one of 1058's inspirations for going 2+2 this year and we absolutely love it, a great blend of speed and pushing power. Spinning around the traction wheel side is actually an upside, if you get T-boned or hit from the traction wheel side spinning out of the pin is very easy, moreso than a standard 6WD.



03-20-2017 07:42 PM

pmattin5459


Unread Re: pic: Subjective Eval of Common Drive Systems

This is nice for those who wouldn't understand a weighted cost-benefit table. However, I don't believe it's a very good substitute for it at all.

First off, it lacks granularity and makes certain drivetrain types seem better than others from a numerical standpoint when they really aren't. Parallel Omnis seems like a great choice for a drivetrain on paper, better than tank drive (hint: it really isn't. I like different drivetrain designs, but go with a 2 x 2 or butterfly or something if you want better turning).

Speaking of tank drive, KOP isn't really the baseline you want it to be. Any decent custom tank drive is more reliable than the KOP chassis, has more pushing power, and more mobility.

To properly explain drivetrain design, it's best to come up with a long-winded post on why tank drive is the best and you should never use anything but tank drive (that's not really what I believe, but it'll go over well on CD). Explain that every single drivetrain other than tank will break down by the end of every match or be so complex that you wouldn't get done with it by the end of build season (again, sarcasm intended, but this is how CD feels).

Seriously though, best way to explain drivetrains is to justify your claims.



03-20-2017 08:12 PM

swootton


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by pmattin5459 View Post
This is nice for those who wouldn't understand a weighted cost-benefit table. However, I don't believe it's a very good substitute for it at all.

First off, it lacks granularity and makes certain drivetrain types seem better than others from a numerical standpoint when they really aren't. Parallel Omnis seems like a great choice for a drivetrain on paper, better than tank drive (hint: it really isn't. I like different drivetrain designs, but go with a 2 x 2 or butterfly or something if you want better turning).

Speaking of tank drive, KOP isn't really the baseline you want it to be. Any decent custom tank drive is more reliable than the KOP chassis, has more pushing power, and more mobility.

To properly explain drivetrain design, it's best to come up with a long-winded post on why tank drive is the best and you should never use anything but tank drive (that's not really what I believe, but it'll go over well on CD). Explain that every single drivetrain other than tank will break down by the end of every match or be so complex that you wouldn't get done with it by the end of build season (again, sarcasm intended, but this is how CD feels).

Seriously though, best way to explain drivetrains is to justify your claims.
So as posted you make general assumptions and imply that most of CD would choose tank drive as best. For many teams it is the best choice for many reason. We are a 6th year team and have run a KOP drive train all 6 years. For recycle rush we took the tread off the center two wheels and ran omnis with an additional pair of omnis in the center, last year we ran aluminum 8" wheels with stock gearing. I would put our bot up against any other bot and I can guarantee that our driver can play hard core defense on you no matter what drive train you have.

We are in New England and it gets rough. We have never broken any drive train component and have gone head to head against 8wd 6 cim, swerve, West Coast, just to name a few. In Waterbury we were pushing not one but two robots across the field at the same time with a 100% KOP drive train.

In Greater Boston there was a second year team that opted out of the KOP and built a swerve drive because the thought the stock drive train wouldn't work well. Because of their budget and team resources they had a very maneuverable robot the could could place 3-4 gears and do nothing else. This was their second event and our team got them climbing by Sunday morning. They didn't do much time and money on building and programming their swerve drive it left no time for any manipulators.

Drive train is just one part of the whole robot. If you don't have the time, money or team resources a fancy drive train is a luxury you can afford if you can't play the game.

If we could we would love to try a different drive train type but until we build, develop and test one before build season it will not be part of our design. We have a simple rule, "If you can't move, you can't play" so until we have another proven design that we can build and program just as quickly with a 100% non-failure guarantee we will be using a KOP drive train because it works.

If you want proof just look for matches at Waterbury or Greater Boston District this year or any event from 2014 that we played defense and you'll have your proof. If we're playing against 558 you can almost certainly see some defense between us and they run 8wd colsons with 6 cims most years.



03-20-2017 08:21 PM

Andrew Schreiber


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by frcguy View Post
I think you're reading it wrong. KoP isn't getting dinged on anything.
This is correct, the KoP is my baseline for everything. From there each attribute is graded as either better, worse, or equal.


Quote:
Originally Posted by Dillon Carey View Post
I'm not sure how your defining power. I'd argue that it's pretty independent from drive train type (except if your accounting for efficiency losses).
For example, you have 2+2 listed as low power. Our 2014 bot had 6 CIMs, which most would argue is plenty (or more) power than you need.
I was more working from staying power which is where it gets dinged because it can be more easily spun.



Quote:
Originally Posted by pmattin5459 View Post
This is nice for those who wouldn't understand a weighted cost-benefit table. However, I don't believe it's a very good substitute for it at all.

First off, it lacks granularity and makes certain drivetrain types seem better than others from a numerical standpoint when they really aren't. Parallel Omnis seems like a great choice for a drivetrain on paper, better than tank drive (hint: it really isn't. I like different drivetrain designs, but go with a 2 x 2 or butterfly or something if you want better turning).

Speaking of tank drive, KOP isn't really the baseline you want it to be. Any decent custom tank drive is more reliable than the KOP chassis, has more pushing power, and more mobility.

To properly explain drivetrain design, it's best to come up with a long-winded post on why tank drive is the best and you should never use anything but tank drive (that's not really what I believe, but it'll go over well on CD). Explain that every single drivetrain other than tank will break down by the end of every match or be so complex that you wouldn't get done with it by the end of build season (again, sarcasm intended, but this is how CD feels).

Seriously though, best way to explain drivetrains is to justify your claims.
You're right, this doesn't replace a good WOT or a good discussion. However, one of the goals was to distill a design/prototyping process to something that can be seen at a glance. Things like this are useful for showing to judges or as documentation.

KoP is a baseline because it's about the only thing every single team can see the performance of. It's a baseline precisely to get everyone on the same page.

Quote:
Parallel Omnis seems like a great choice for a drivetrain on paper, better than tank drive
I'm not sure if you're quite reading this chart properly. On paper Asteroid Drive (I'm trying Nate!) is about equal to the KoP drive, it's a bit more mobile but has less staying power.



03-20-2017 08:49 PM

marshall


Unread Re: pic: Subjective Eval of Common Drive Systems

Zebracorn Swerve - Taking Unicorn Swerve and making it more complicated with ROS, Stereocameras, and White Papers.



03-20-2017 08:55 PM

Andrew Schreiber


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by marshall View Post
Zebracorn Swerve - Taking Unicorn Swerve and making it more complicated with ROS, Stereocameras, and White Papers.
Just for clarification on where 900's drivetrains fall:



03-20-2017 08:59 PM

PayneTrain


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Andrew Schreiber View Post
Just for clarification on where 900's drivetrains fall:

hey now, those harpoons are undefeated



03-20-2017 10:51 PM

pmattin5459


Unread Re: pic: Subjective Eval of Common Drive Systems

Here's my drivetrain philosophy has 3 Sacred Rules, in order of most importance to least:

1. Build what you know. If you've never built a drivetrain type before, avoid it. Nearly every team knows how to build a tank drive and every team should be able to build a KOP drivetrain, so naturally these are the most common and well-loved drivetrains out there. But there are teams that excel using non-tank drivetrains that are more reliable than some tank drives. Why? Because they've gone over several design iterations over years of competition (see 148's Nonadrive for a good example, or 16's legendary swerve).

2. Build what fits. Weight and space restrictions. Pretty obvious, but it needed to go somewhere.

3. Build what fits with your strategy. That's where weighted tables and such come in handy. If you have multiple drivetrains that fit the first two rules, this is how you pick the right one. But any candidates at this point must satisfy the above rules.

I think this is a more precise and nuanced answer to "what drivetrain should we use" than simply "tank drive" or "depends on the game".



03-20-2017 11:47 PM

Andrew Schreiber


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by pmattin5459 View Post
Here's my drivetrain philosophy has 3 Sacred Rules, in order of most importance to least:

1. Build what you know. If you've never built a drivetrain type before, avoid it. Nearly every team knows how to build a tank drive and every team should be able to build a KOP drivetrain, so naturally these are the most common and well-loved drivetrains out there. But there are teams that excel using non-tank drivetrains that are more reliable than some tank drives. Why? Because they've gone over several design iterations over years of competition (see 148's Nonadrive for a good example, or 16's legendary swerve).

2. Build what fits. Weight and space restrictions. Pretty obvious, but it needed to go somewhere.

3. Build what fits with your strategy. That's where weighted tables and such come in handy. If you have multiple drivetrains that fit the first two rules, this is how you pick the right one. But any candidates at this point must satisfy the above rules.

I think this is a more precise and nuanced answer to "what drivetrain should we use" than simply "tank drive" or "depends on the game".
I still think you're mistaking the point - I don't CARE what choice teams make. I want them to start putting things in easily digestible chunks for judges, sponsors, younger students. But because representing data is hard I wanted to share a viable way to do this that most folks could create in Powerpoint if needed.

Unrelated.- it's ironic you mention 16's legendary swerve as this came out of a comparison of their drivetrain with JT :



03-20-2017 11:55 PM

marshall


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Andrew Schreiber View Post
I still think you're mistaking the point - I don't CARE what choice teams make. I want them to start putting things in easily digestible chunks for judges, sponsors, younger students. But because representing data is hard I wanted to share a viable way to do this that most folks could create in Powerpoint if needed.

Unrelated.- it's ironic you mention 16's legendary swerve as this came out of a comparison of their drivetrain with JT :
I feel as though this point is lost on many who would rather have a debate about the merits of a holy war... waiting on someone to show up and start going on about Endians and eggs now.



03-20-2017 11:59 PM

Andrew Schreiber


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by marshall View Post
I feel as though this point is lost on many who would rather have a debate about the merits of a holy war... waiting on someone to show up and start going on about Endians and eggs now.
Holy Wars are an effective way to empty orphanages



03-21-2017 12:35 AM

Mechvet


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by marshall View Post
Zebracorn Swerve - Taking Unicorn Swerve and making it more complicated with ROS, Stereocameras, and White Papers.
Wait, you guys use ROS? Happen to have a link where I can creep on your integration?



03-21-2017 12:43 AM

marshall


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Mechvet View Post
Wait, you guys use ROS? Happen to have a link where I can creep on your integration?
Yes and when build season is over we'll do our thing that we do. At the moment, if you're really interested then look at what Tuft's is doing with Baxter and ROS for LabVIEW. We borrowed (READ: stole) heavily from their work.



03-21-2017 07:43 AM

Taylor


Unread Re: pic: Subjective Eval of Common Drive Systems

It's cute how you label it as "subjective" so you don't have to entertain opposing views.

But really it's not about drive trains. It's about decision making and documentation and organization and clear communication. To that end, I think this works well.



03-21-2017 07:58 AM

KJaget


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Andrew Schreiber View Post
Just for clarification on where 900's drivetrains fall:

I expected more red.



03-21-2017 07:58 AM

Andrew Schreiber


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Taylor View Post
It's cute how you label it as "subjective" so you don't have to entertain opposing views.

But really it's not about drive trains. It's about decision making and documentation and organization and clear communication. To that end, I think this works well.
Yup, I could have made it about types of cookies (which oatmeal raisin is the definitive loser, fight me) but figured making it something at least somewhat pertinent might be of more use.



03-21-2017 08:00 AM

marshall


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by KJaget View Post
I expected more red.
It's actually a graph of my energy levels the day we built it between caffeine runs.



03-21-2017 08:03 AM

Chris is me


Unread Re: pic: Subjective Eval of Common Drive Systems

I think the arrows in the graph make it confusing. "Weight, up, good thing?". Just make it red and green dots, maybe with + or - in them if you need some indicator of good / bad for colorblind people.

2+2 drive is pretty dope, but you'll be ripping my 6WD away from my cold, dead hands.



03-21-2017 08:37 AM

JesseK


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Chris is me View Post
2+2 drive is pretty dope, but you'll be ripping my 6WD away from my cold, dead hands.
Another column could be 'Practice Needed', which represents how much practice the driver needs with the drive train in order to maximize its effectiveness. All of the holonomics (except maybe 'H') take a hit there, as does 2+2.



03-21-2017 08:55 AM

TimTheGreat


Unread Re: pic: Subjective Eval of Common Drive Systems

What exactly does 'Software' mean. You have mecanum listed as being harder than KoP, when it's changing

Code:
robotDrive.tankDrive(left, right)
to
Code:
robotDrive.mecanumDrive_Cartesian(y, x, rot)



03-21-2017 08:56 AM

Chris is me


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by TimTheGreat View Post
What exactly does 'Software' mean. You have mecanum listed as being harder than KoP, when it's changing
Code:
robotDrive.tankDrive(left, right)
to
Code:
robotDrive.mecanumDrive_Cartesian(y, x, rot)
To put it simply, dead reckoning code for mecanum drive isn't good enough to use effectively in competition, whereas dead reckoning tank drive code is.



03-21-2017 08:59 AM

Andrew Schreiber


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by TimTheGreat View Post
What exactly does 'Software' mean. You have mecanum listed as being harder than KoP, when it's changing
Code:
robotDrive.tankDrive(left, right)
to
Code:
robotDrive.mecanumDrive_Cartesian(y, x, rot)
Sorry, I meant "Software required to be effective".

You can absolutely do that but it will be harder to drive - you'd be more effective adding gyros and encoders in to get more consistent behavior (yes in tele-op).



03-21-2017 09:10 AM

Oblarg


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by JesseK View Post
Another column could be 'Practice Needed', which represents how much practice the driver needs with the drive train in order to maximize its effectiveness. All of the holonomics (except maybe 'H') take a hit there, as does 2+2.
+1



03-21-2017 09:12 AM

marshall


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by Andrew Schreiber View Post
Sorry, I meant "Software required to be effective".

You can absolutely do that but it will be harder to drive - you'd be more effective adding gyros and encoders in to get more consistent behavior (yes in tele-op).
A discussion with another team recently told me something interesting about our team... It occurred to me in the course of the conversation about encoders and drivetrains that our code is 100% reliant on them. We couldn't do any of the cool stuff we're doing without encoders.



03-28-2017 09:57 PM

messer5740


Unread Re: pic: Subjective Eval of Common Drive Systems

Question about the 8wheel drive train where the outside wheels are Omni, do you need to include a drop for the center wheels? I'm pretty sure you don't but I'm just making sure.



03-28-2017 10:39 PM

Oblarg


Unread Re: pic: Subjective Eval of Common Drive Systems

Quote:
Originally Posted by marshall View Post
A discussion with another team recently told me something interesting about our team... It occurred to me in the course of the conversation about encoders and drivetrains that our code is 100% reliant on them. We couldn't do any of the cool stuff we're doing without encoders.
We've moved in this direction over the past two years, and are never going back. Velocity-controlled drives handle so much better than open-loop voltage-controlled drives, and dead reckoning autonomous without feedback is atrocious.



view entire thread

Reply
previous
next

Tags

loading ...



All times are GMT -5. The time now is 04:18 PM.

The Chief Delphi Forums are sponsored by Innovation First International, Inc.


Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi