Actuated Drive (Slide / H)

[GeeTwo]

Our 2015 robot was designed to have an H/slide drive, with a fixed altitude center wheel (we did not plan to drive onto or over the scoring platforms). We never did get strafe working well, and removed it from the robot during competition. As one of our fall training exercises, we are going to attempt rebuilding our practice chassis as a pneumatically-actuated H drive. To simulate a competition robot, the whole thing including battery and bumpers will weigh about 140 lb, even if that means adding ballast. The chassis is a 2015 KoP frame, long configuration (31" long, 25" wide), and we'll use AM DuraOmni 4" wheels on the corners. We stripped the tread off of two of the KoP wheels to make a spacer on the drive axle so that we have a 4 wheel drive before adding strafe. We already have some 8.45:1 Toughbox2 gearboxes with the long hex shaft, so that we will have the same gear ratio for the strafe as we're using for the tank drive. This is probably not important for this discussion, but we intend to make bumpers and an active (8-9") ball pickup as part of the prep for the 2016 game as less than a fifth of our current team was with us in 2014.

Now, to the questions:

• This is our first actuated drive; what general lessons learned or pitfalls to avoid with actuated drives can you give us?
• My instincts want to start with about a 2" throw on the wheels (with carpet contact being near the center of the throw), and between 1/4 and 1/2 of the robot weight (nominally 1/3) being supported by the strafe wheel(s) when the cylinder(s) force them down. Are these numbers about right, too high, or too low for motion on level carpet?
• How much advantage is there to doing multiple strafe wheels? Is it a serious problem if a single strafe wheel is a little bit (no more than 2") off of the center of mass?
• Any preference for the hinge/pneumatic axis to be parallel or perpendicular to the wheel axis?
• Any other bits of help, encouragement, or discouragement from this line of attack?

OBTW, I do plan for the strafe wheel contact patch to be directly between the hinge(s) and the cylinder(s) to minimize stray torques.

[Abhishek R]

The strafe wheel needs to be centered relative to the mass of the robot rather than the geometry. If for some reason this is not possible due to other mechanisms, structural support, etc. you will need some other way to correct for the robot's curved movement when attempting to strafe (like a gyro assisted correction system) since it probably won't drive in a straight line.

[GeeTwo]

Originally Posted by Abhishek R

The strafe wheel needs to be centered relative to the mass of the robot rather than the geometry. If for some reason this is not possible due to other mechanisms, structural support, etc. you will need some other way to correct for the robot's curved movement when attempting to strafe (like a gyro assisted correction system) since it probably won't drive in a straight line.

Yes, I had a detail question which I would not have asked if I did not know this:

Originally Posted by GeeTwo

• How much advantage is there to doing multiple strafe wheels? Is it a serious problem if a single strafe wheel is a little bit (no more than 2") off of the center of mass?

Unfortunately, after the mechanicals were done on this, we shifted to doing an active intake, lift, and dumper for tennis balls. One works well (intake), one fairly (dumper), and one poorly (lift, based on velcro gripping the ball). The team was working on this one through yesterday evening; now we're taking two weeks off for the Christmas/New Year holiday, and will come back doing mostly mental and organizational prep for kickoff.

[Owen Busler]

Here you can see our 2015 robot with a pneumatically actuated h wheel. The original reason for it to be actuated was so we could drive over the scoring platforms. But, once we began testing we found out the 6" vex pro omni would just slip. We replaced the omni with a traction wheel and it works beautifully.

https://youtu.be/6lu_WeM54G0?t=41s

The wheel is slightly in front of our center of mass, so we would go in a circle when we strafed. Our programming team coded some very nice correction software which used encoders and the navX to make the bot strafe perfectly sideways.

[GeeTwo]

Originally Posted by Owen Busler

Here you can see our 2015 robot with a pneumatically actuated h wheel. The original reason for it to be actuated was so we could drive over the scoring platforms. But, once we began testing we found out the 6" vex pro omni would just slip. We replaced the omni with a traction wheel and it works beautifully.

https://youtu.be/6lu_WeM54G0?t=41s

The wheel in the video is an omni, and it looks OK there. Did you mean perhaps that the omni slipped when attempting to strafe on the scoring platform?

Do you recall how much weight you put on the strafe wheel (pressure times area times ratio of lever arms)?

[notmattlythgoe]

Originally Posted by GeeTwo

The wheel in the video is an omni, and it looks OK there. Did you mean perhaps that the omni slipped when attempting to strafe on the scoring platform?

Do you recall how much weight you put on the strafe wheel (pressure times area times ratio of lever arms)?

There also doesn't appear to be any game pieces on the robot. Add a stack of totes and I can see where the issues would come from.

[Owen Busler]

Yes thank you for catching that. When we had a stack the omni would slip on the carpet, which is why we switched it out for a traction wheel.

These are my basic calculations:

1.25 inch cylinder
.4 inch in diameter rod inside cylinder
60 psi

(1.25)(pi)= 3.92 square inches
3.92-(.4*pi)=2.67 net square inches
2.67*60=160 lbs

The pivot point was equidistant from where the piston was mounted and the center of the wheel.

[GeeTwo]

Originally Posted by Owen Busler

Yes thank you for catching that. When we had a stack the omni would slip on the carpet, which is why we switched it out for a traction wheel.

These are my basic calculations:

1.25 inch cylinder
.4 inch in diameter rod inside cylinder
60 psi

(1.25)(pi)= 3.92 square inches
3.92-(.4*pi)=2.67 net square inches
2.67*60=160 lbs

The pivot point was equidistant from where the piston was mounted and the center of the wheel.

The area of a circle is πr², not 2πr. This makes the area π(1.25² - 0.4²)/4 = 1.1 in² net, or 66 pounds. If you'd pushed down with 160 pounds, the robot would have been lifted off of half of the edge wheels, and you should have had enough traction even with a tall stack of totes.

As I understand this both from your last sentence and the subtraction of the shaft, you had the cylinder pulling in order to push the wheel into the carpet; is this correct? Even though the difference isn't usually that large, we've been intentionally trying to load our pneumatics so that the greatest needed force is delivered when the shaft is moving outward; if we'd done that with our arms in 2014, we probably would have saved ourselves some headaches at competition.

[Owen Busler]

Geez i must be out of it today. Thanks for correcting that. It did seem way too high.

But yes what you have stated is correct.

[DaveL]

Originally Posted by Owen Busler

The wheel is slightly in front of our center of mass, so we would go in a circle when we strafed. Our programming team coded some very nice correction software to make the bot strafe perfectly sideways.

+1

On our prototype bot we mounted the strafe wheel to a transverse support, like the trailing arm in a front wheel drive car. We used a week spring of approx 1/5 of the weight of the bot on a 4 wheel drive bot.

Going in sideways in one direction worked better. It seemed to dig in better. Going the other sideways direction the wheel seemed to ride up and slide more. We ended up with a vertical slide spring, but I would have rather used the traverse arm with pneumatics as I believe it would be lighter and then we could adjust the downward force and lift it up when not needed.

Did anyone build this type of mechanism and use pneumatics to deploy it?
Dave

[dradel]

Our team used an actuated H last year and it worked without any hiccups. One thing I would recommend is make it so there is some adjustment available to raise or lower the center wheels to get more or less pressure on the floor surface.

[GeeTwo]

Originally Posted by DaveL

Going in sideways in one direction worked better. It seemed to dig in better. Going the other sideways direction the wheel seemed to ride up and slide more.

What type and orientation of gearing did you use? Perhaps the drive gear was putting a lift force on the wheel in one direction, and a depression in the other. This is most likely to happen with a horizontally oriented gear train (such as most versablock setups). Using a planetary or vertically oriented gearbox (such as how the TB-mini is mounted on the KoP chassis) might solve the problem.

[DaveL]

Originally Posted by dradel

Our team used an actuated H last year and it worked without any hiccups.

Nice setup!

The trailing arm that I first designed looked like the attached diagram.

I'm thinking if I replaced the spring (green texture) with a cylinder, the structure would have been more rigid and the wheel would not have lifted when moving left.

The other option would be to raise and lower a larger structure like dradel's post.

Dave

[Chris Hibner]

We did an H-drive in 2011 with a lot of success (won a district, made it to Einstein). We did not actuate it, but it was on a swing arm and loaded with a gas strut. We sized the gas strut such that in the normal position it applied a force of approximately 1/5 of the robot weight on sideways wheel.