pic: Prototype Drivetrain

[Akash Rastogi]

Originally Posted by 548swimmer

You must not have understood what I meant. With this drivetrain design and current wheel placement, it will never be terribly easy to control. With this in mind, I figured it would not make it too much worse to pull the wheels in. We also would be able to use a gyro to help keep it straight (we used one for autonomous this year).

....which would defeat the purpose of this drive.

With this in mind, what are you essentially gaining? Just trying to find the reasoning for this since the wider stance of the wheels would be better for turning and control. What is the purpose you are trying to get out of this?

.

[548swimmer]

Originally Posted by LLogan

I would advise you to drive the four omnis on the corners. Your tractive force is based off of the sum of the forces of friction on all of your wheels. With the design you have right now, you are essentially "wasting" the normal force on the corner wheels. Whatever weight you will have on those sections of the robot will not contribute to your pushing force, giving you less than what you would generally want.

Though the force is "wasted" the majority of the weight will be placed on the drive wheels due to where the CG is. Our current build leader, whom I have discussed this design with, guesstimates that roughly 75% of the weight will be put on the center wheels. Also, I believe that in previous posts I addressed the design goal of maneuverability. The omni's are there as casters, meant to prevent the robot from scratching the floors. There is also the suspension, that will transfer weight during a pushing match onto the center wheel.

[548swimmer]

Originally Posted by Akash Rastogi

With this in mind, what are you essentially gaining? Just trying to find the reasoning for this since the wider stance of the wheels would be better for turning and control. What is the purpose you are trying to get out of this?

.

You gain the ability to have a zero turn radius, and turn really quickly. Anticipating lack of control even with a "normal" wheel configuration, I opted for a tighter center system. Programming can keep us straight. I have faith in them . Additionally, we can just turn down the sensitivity of the joysticks until they reach 50% of their total travel, allowing finesse, while still maintaining your ability to react quickly.

[Chris is me]

Originally Posted by 548swimmer

Anticipating lack of control even with a "normal" wheel configuration, I opted for a tighter center system. Programming can keep us straight. I have faith in them .

So you saw there might be a problem, and decided to make it worse... why?

[548swimmer]

Originally Posted by Chris is me

So you saw there might be a problem, and decided to make it worse... why?

Because it CAN be done. This is just a concept. Teams do just fine with mecanums or all omni's. Neither of these designs provide much pushing power. I do not want this robot to be driven just to push stuff around. My goal is to inspire those below me who saw our robot from this year push someone accross a gym floor while they were standing up, or tow 3 people at once. I want this robot to be driven with a unique drive philosophy, and the best way to do that is make it a unique robot.

If this isn't used, but gives one of the design team a new idea to help accomplish whatever we will be challenged with, it will be a success in my mind.

[Jack Jones]

Programming can keep us straight. I have faith in them

Physics is not a religion.

[Akash Rastogi]

Originally Posted by Jack Jones

Physics is not a religion.

Hahahah.

To help elaborate on what Chris, apalrd, and I were talking about, I urge you to read "Drive Train Basics" by Chris Hibner and the accompanying drivetrain calculator that goes along with it created by Mark Kramarczyk. Use your CAD to plug in your distances into the calculator.

This may prevent you from making somewhat of a very silly and obvious mistake.
Drive Train Basics: How to be sure your robot will turn
Turning Force Analysis

Please at least read through Chris Hibner's paper as it includes the physics that you need to look over. Pay close attention to changes in LsubTW

For some reason I have a feeling you're confusing the difference between the advantage of a "short wheel base" (wheel to wheel distance) and the distance between parallel wheels. Once you go through the pdf you'll see what I mean. Have fun.

PS- Non driven wheels are wasted weight and wasted force. Just fyi. I'm still not understanding the rationale you're giving but good luck! Nice CAD.

[548swimmer]

Originally Posted by Jack Jones

Physics is not a religion.

No, though it does apply to robots. With the simple push of a button, a gyro sensor could be enabled(wrong word) allowing only forward and backward movement. Or you could go with arcade drive, and as long as the joystick stays 1 degree from the y-axis, you only allow straight movement. That number could be tweaked untill it's optimized.

[548swimmer]

Originally Posted by Jack Jones

Physics is not a religion.

Originally Posted by Akash Rastogi

Hahahah.

Please at least read through Chris Hibner's paper as it includes the physics that you need to look over.

For some reason I have a feeling you're confusing the difference between the advantage of a "short wheel base" (wheel to wheel distance) and the distance between parallel wheels. Once you go through the pdf you'll see what I mean. Have fun.

PS- Non driven wheels are wasted weight and wasted force. Just fyi. I'm still not understanding the rationale you're giving but good luck! Nice CAD.

I read through the paper, and I discussed all if that today with our design leader. I understand what makes/prevents a robot from turning. No matter where these wheels are, the robot WILL turn. This design may not always be useful, but both he and I agree that it is worth investigating. If you approach the driving of this robot as a robot to be handled with a gentle touch instead of intense fast paced and borderline reckless driving that occurs in some games, you might see where I'm coming from.

[Dave McLaughlin]

Just to clarify, you are engineering for maneuverability?

[LLogan]

Originally Posted by 548swimmer

Though the force is "wasted" the majority of the weight will be placed on the drive wheels due to where the CG is. Our current build leader, whom I have discussed this design with, guesstimates that roughly 75% of the weight will be put on the center wheels. Also, I believe that in previous posts I addressed the design goal of maneuverability. The omni's are there as casters, meant to prevent the robot from scratching the floors. There is also the suspension, that will transfer weight during a pushing match onto the center wheel.

At the cost of 25% of your tractive force, you gain a theoretical slight advantage in turning. This should raise a very large red-flag. I don't think that losing 25% of your pushing force has ever been a good thing. The turn radius for a dropped center 6WD isn't zero, but it can be pretty close. Driving the omni wheels will not cost you much maneuverability at all. The suspension will not magically "transfer" weight to the center. In fact, a suspension is useful when you want to keep all of your wheels on the ground. A suspension makes sense in a mecanum drive, where you want to make sure that the normal force is distributed as evenly as possible, otherwise the perpendicular vectors don't cancel each other out and you cannot drive straight. The springs in your suspension will actually cause a reactive force and increase the normal force on your omnis, which is the exact opposite of what you want to happen.

I apologize if I'm making the assumption that your omni wheels are on an active suspension. But... I would be even more puzzled if you were to use a floating suspension.

[artdutra04]

Originally Posted by 548swimmer

No, though it does apply to robots. With the simple push of a button, a gyro sensor could be enabled(wrong word) allowing only forward and backward movement. Or you could go with arcade drive, and as long as the joystick stays 1 degree from the y-axis, you only allow straight movement. That number could be tweaked untill it's optimized.

Don't be so quick to assume everything in code will be worked out flawlessly.

When you design mechanically unstable systems, you better have absolute 100% reliable software, otherwise you're a sitting duck.

Putting the main drive wheels right next to each other isn't a smart idea unless you want to implement real-time, closed-loop PD velocity controllers on your drive train. Regardless of physics, putting the drive wheels right next to each other is a bad idea geometrically. Let's assume that since not all motors have identical RPMs (whether due to motor bias or manufacturing tolerances), that one side of your drive train normally operates 5% faster than the other one.

To minimize risk of unpredictably turning with slightly different wheel speeds, you should keep the wheels as far apart as possible. This way, the ICR (instantaneous center of rotation) will be moved much further away from the center of your robot as opposed to a smaller wheelbase. As it is, most existing 6WD drivetrains already turn way too fast at full power to be accurately controllable, with many teams already resorting to smoothing, ramping, or square/cubic response curves.

[synth3tk]

I may not be a mechanical/physics person so I have no idea what's being discussed here, but I do know that relying on the programmers to fix something that's wrong with mechanics is never a good idea, especially if the issue can be avoided ahead of time. As Mr. Dutra said, you'd better have flawless programming to back up the design downfalls.

[Edoc'sil]

The others were more polite about this, I am not. Wheels in the center like this are dumb. In January we build a DT like this, just with the wheels in the logical positions and all of them powered. Surprise! It has 0 turning radius as well! It also was almost imposible to have any fine control of the movement. After about 15 hours of testing I was still the only one who could drive the thing. Currently we are modifying our DT with a lower gear ration to gain more precision, and a shifter to maintain speed.

If you are set on this idea, build it during the off season and try it out, I am all for learning. But please, don't build this during those 6 tense weeks, you team will murder you.


Bluntly, Edoc'sil

[sgreco]

Originally Posted by Edoc'sil

The others were more polite about this, I am not. Wheels in the center like this are dumb. In January we build a DT like this, just with the wheels in the logical positions and all of them powered. Surprise! It has 0 turning radius as well! It also was almost imposible to have any fine control of the movement. After about 15 hours of testing I was still the only one who could drive the thing. Currently we are modifying our DT with a lower gear ration to gain more precision, and a shifter to maintain speed.

If you are set on this idea, build it during the off season and try it out, I am all for learning. But please, don't build this during those 6 tense weeks, you team will murder you.


Bluntly, Edoc'sil

Just to add to this, it seems that a main reason behind this drivetrain is a zero turn radius. Take a look at some traditional 6 wheel skids in video, they all have zero turn radii as well, except without the loss of control.

08 was the last year we had basic drivetrain (bumps changed things this year and regolith did last year), watch teams like 968/254, 1114, 217, 1717, 330. These bots were incredibly manuverable and they all had a zero turn radius.

Edit: Check out this thread. http://www.chiefdelphi.com/forums/sh...threadid=85710, watch the video...I know its an 8 wheel, but the concept is the same. All of the wheels are on the outside and they have no problem with manueverability.