Treads, Tracks, Tank system Talk/Help

[UndergroundVoid]

The Background:
Alright, I would like to say to start off I love treads/ tracks/tank drive and feel the need to please the tank gods. My team has had a tank drive system for one year and support after that year for tank type drive systems have died and have be replaced with others systems aka wheels, omnis, mecanum, etc. The robot equipped with the tank drive system was for a while my roommate as i had made repairs but found out why then bot had so much issues like fractured lex sand framework and really bad gearboxes but yet it holds true to my heart. I'm revamping its design which is near completion but has some questions that need answering before it can be completed and also it may have more questions that i may have not thought of.

The Question:
1. What gear ratios should i have for the tank tank system (motor to gearbox output)?
Note: I'm not stupid i know there is a series of equations to find the right ratio but I'm looking mainly for what others have done that might work for me.
2. How to solve the walking effect where i seen some robots with treads hobble as they turn and even become unbalanced that the driver has to stop?
I assume this is from the tracks not being held center because of certain belt track systems.
3. How to correctly tension the tank system, such as spring loaded or motor controller tension or to have the tracks at a fixed tension set by the frame?

General Talk:
I'm also looking for just a general what you may have done in this field of drive train and certain things you may have experienced that may help me and everyone have a better idea on how to produce tank systems to please the tank gods.

[R.C.]

997 has a very impressive tank drive, might want to ask them.

[Akash Rastogi]

as do the RoboWizards 522. Feel free to contact any of their active CD members.

[Al Skierkiewicz]

Some of the issues you are asking about are typical of tank tread systems. One that you have not mentioned is the bane of tank drive using treads or locked wheels, i.e.stalled motors.
When using treads on carpet, the side friction (during turns) with the carpet is so high that robot has to hop to break free from the carpet surface. Since the motor/transmission is usually capable of enough torque to overcome the friction, the robot hops when turning. However, during these moves the motors are at, or very near, stall. In a four motor drive using the CIMs, that is potentially over 400 amps. In the IFI control system, this causes the battery voltage to temporarily fall below the drop out voltage of 8 volts, and the controller goes into protect mode disabling all outputs and switching to the backup battery. (We don't have any data yet on what the new controllers might do under these conditions.) A wheeled robot in the same kind of drive could have a little less side friction but the problem remains and they also will hop to some extent in turns. Drivers can minimize the hopping by turning in conjunction with forward or backward motion. The AndyMark style wheels that have no side friction were developed for this reason. Everyone tries tank treads from time to time, but the tradeoffs are usually significant enough that teams go back to wheeled drive systems. Any game in which turning is required will have a minimum number of robots with treads. This year's game is a good example.

[IKE]

I have no robot experience with tank treads, but most tractors used a single powersource into a differential with turning brakes. This is common on wheeled tractors too (they are often called plow brakes). Using the differential and side brakes essentially allows for torque vectoring and keeping from a stall condition. A couple of teams did this style of drive system this year to help with Car-style steering robots. This probably won't help with the side traction hopping, but it may help keep the current loads lower.

[JesseK]

Originally Posted by IKE

I have no robot experience with tank treads, but most tractors used a single powersource into a differential with turning brakes. This is common on wheeled tractors too (they are often called plow brakes). Using the differential and side brakes essentially allows for torque vectoring and keeping from a stall condition. A couple of teams did this style of drive system this year to help with Car-style steering robots. This probably won't help with the side traction hopping, but it may help keep the current loads lower.

I watched a couple of shows on Discovery (one was JYW where their task was to build a tank) and I believe actual tanks used to be built the same way.

An idea I played around in CAD a few months ago took the 6WD rocking-style drive train and converted the front four wheels into tank treads. The design put most of the weight in the back so the 'bot never really rocked forward onto the treads, however the tank treads would greatly improve the ability to climb ramps. Maybe it's worth exploring, maybe not...I didn't have time to fully CAD it out and consider the possibilities.

[AndyB]

You can greatly increase the mobility of the tread system if you decrease the surface contact.

Take team 522 for example. Year after year they produce a successful tread system.

http://www.chiefdelphi.com/media/photos/17294

Their surface contact basically creates a square profile on the ground and allows the robot to turn with more ease.

[Gdeaver]

Thinking about tank treads and other drive configurations is probably a good idea. The past few years many teams have gone to 6 small wheel designs. For the most part the games have been played on flat surfaces. This has not always been true. What would your team do for a drive system if this years game required going up stairs or getting up on a platform?

[Cory]

Originally Posted by Gdeaver

Thinking about tank treads and other drive configurations is probably a good idea. The past few years many teams have gone to 6 small wheel designs. For the most part the games have been played on flat surfaces. This has not always been true. What would your team do for a drive system if this years game required going up stairs or getting up on a platform?

Pull yourself up to it from the floor

[Alan Anderson]

Originally Posted by Gdeaver

What would your team do for a drive system if this years game required going up stairs or getting up on a platform?

I'd suggest triwheels.

But I'm just a programmer, so when it comes to mechanical design I don't get too vocal about my preferences.

[Dick Linn]

Originally Posted by Gdeaver

... What would your team do for a drive system if this years game required going up stairs or getting up on a platform?

We did that once with a 2-wheel-drive setup. It all depends on the specifics.

[=Martin=Taylor=]

Originally Posted by Gdeaver

Thinking about tank treads and other drive configurations is probably a good idea. The past few years many teams have gone to 6 small wheel designs. For the most part the games have been played on flat surfaces. This has not always been true. What would your team do for a drive system if this years game required going up stairs or getting up on a platform?

Has there been a game yet that REQUIRED teams to climb obstacles? Seems a lot of the successful teams from 1999 and 2004 found clever ways to avoid climbing things.

[AndyB]

Originally Posted by Hachiban VIII

Has there been a game yet that REQUIRED teams to climb obstacles? Seems a lot of the successful teams from 1999 and 2004 found clever ways to avoid climbing things.

There have been games with obstacles that needed to be climbed. But I understand what your saying.

But treads, WHEN DONE RIGHT, are a great option for maneuvering up and over obstacles such as steps, ramps, etc...

There is a reason why tanks use treads.

[Andrew Schreiber]

Originally Posted by AndyB

There is a reason why tanks use treads.

Yes, the reason is that they are better at spreading the weight of an 80 ton battle tank when crossing mud.

I won't advise you how to do tank treads because I don't know how, but I will advise you to look to the military and their move away from treads. (run a search on Stryker) Tank treads are great for uneven and shifting terrains but for speed and maneuverability they are not the most effective.

Now, would it be cool to do a tank tread system that could climb a 70 degree incline as a demo? Oh you bet it would be. My best suggestion would be to look at the military tanks and how they work. An interesting thing to do might be to have a way of letting your treads lose friction as they are moving sideways (similar to omni wheels) Might be interesting to make work, dont know how useful it would be though.

Just my (programmer's) $.02

[Pat Roche]

Originally Posted by Al Skierkiewicz

Some of the issues you are asking about are typical of tank tread systems. One that you have not mentioned is the bane of tank drive using treads or locked wheels, i.e.stalled motors.
When using treads on carpet, the side friction (during turns) with the carpet is so high that robot has to hop to break free from the carpet surface. Since the motor/transmission is usually capable of enough torque to overcome the friction, the robot hops when turning. However, during these moves the motors are at, or very near, stall. In a four motor drive using the CIMs, that is potentially over 400 amps. In the IFI control system, this causes the battery voltage to temporarily fall below the drop out voltage of 8 volts, and the controller goes into protect mode disabling all outputs and switching to the backup battery. (We don't have any data yet on what the new controllers might do under these conditions.) A wheeled robot in the same kind of drive could have a little less side friction but the problem remains and they also will hop to some extent in turns. Drivers can minimize the hopping by turning in conjunction with forward or backward motion. The AndyMark style wheels that have no side friction were developed for this reason. Everyone tries tank treads from time to time, but the tradeoffs are usually significant enough that teams go back to wheeled drive systems. Any game in which turning is required will have a minimum number of robots with treads. This year's game is a good example.

Your center of gravity has a very impact on this issue also. Assume you have a four cog setup (three similar to a six wheel drive setup and one tensioner) and your weight is either towards the front cog or the back cog you will run into this issue. However by finding the happy medium between center cog drop and your center of gravity as close to the center/low as possible many track systems level out on that center wheel as they reach. My understanding is that this is due to the tread actually sinking a little. This gives the robot a slightly larger footprint to balance on. Account for that and the length of the belt touching the carpet,its like turning a very short but wide wheel base. Another thing that drivers tend to do is not let the robot settle for a second before trying to turn this often throws off the center of gravity way over to the front/rear cogs. To settle the robot you don't necessarily have to stop the machine but just let up slightly to bring down the robots acceleration and allow the robot to tilt back on plane with the center cogs. However I do agree with Al in saying if it is not necessary for the game why implement it. Tread's are superb and often faster at vertical obstacles(stairs, bumps, etc.) but in a flat floor game it tends to be prohibitive.

Best of luck,

Pat