The other evening I was experimenting with the new Tank Tread Upgrade Kit, putting one traction link in for each two regular (green) links. Then I started thinking: Wouldn’t it be cool if EVERY segment was a traction link!
That lead to the question: What performance improvement do you get on various surfaces as you replace more links with traction links?
So now I’m considering putting together a full blown experiment where I try various combinations of traction link arrangements on various surfaces. I’d build a well-balanced, low-CG, torquey “crawler” bot as the test rig.
I figure I would do a “towing strength” test on level ground using a force gauge, and I could do some kind of “holding the slope” test where I increase the angle of a ramp until each tread configuration can’t make it up. I also thought I could test the effects on turning speed. Then write up all my data for the world to share and enjoy.
Questions:
[LIST=1]
*]Has anybody seen such a traction study done with Vex already?
*]Would folks on the forum generally find this info useful?
*]What particular tests are of interest?
*]Which surfaces should I use? (Concrete, plywood, carpet,…)
*]Should I also include the standard wheels, or just do tank tracks?
*]Any particular ideas regarding methodology?
[/LIST]
Hey everybody,
Thanks for the feedback, both on the forum and via PMs. I’m going to be writing up my basic approach and posting it shortly. I am trying to do this up right, so it may take a little time to pull it all together.
To measure the towing force accurately, I plan to buy a small force gauge. The good ones with serial ports to capture the data start at $500 and go up from there :eek:. Here is an example of one I’m looking at.
I have always wanted a decent force gauge, so if you have experience with a particular model or dealer, please post your opinions (good or bad). My ideal gauge would be one that could be adapted to measure both linear force (compressive and tensile) as well as rotary (CW & CCW), though I’ve not seen one of those for less than $1000…
Using a force gauge sounds really cool but do you really have $500 to get one?
Who said it had to be digital, here is another way to measure the force.
Order a Digital scale and some ball bearings or small weights.
Now build a squarbot with no motors, just the frame and wheels.
Now build some kind of bucket to hold the weight and then measure that weight.
Once you are done you can measure out the weight of each piece that you put in uthe until vex cannot pull it anymore. Then just find the total weight of the square bot and that should give you its force.
I don’t know if this would be accurate because the frame in the back is on wheels which minimizes friction. If you took off the wheels you would increase the friction which might give you more accurate results.
But if you can afford the digital force tester than get that.
Check this out, i found a good deal on a force tester on ebay so always check ebay.
Well, I’m not crazy about the idea of dropping that much coin on a force gauge, but I also don’t mind investing in good tools. I’ve built up a fairly well-equipped shop over the years by buying the best tools I could afford, one at a time, as projects demanded them.
Yep - it did occur to me to build something. My first thought was to build a force gauge out of Vex!!! A sliding mechanism with a spring, and a potentiometer on a rack-and-pinion to measure how far the spring extends. I have no idea how accurate/repeatable it would be, though.
Thats a neat idea, and I might just do something like that if I don’t spring for a digital gauge.
Nice - that certainly helps a bit!
Also, I’m going to see if the TechShop near me has one I can get access to.
I don’t know. Seems like the average force would be a better indication of actual traction. With peak force, all it would take is one brief moment of extra traction to provide an unrealistic result.
On the other hand, $30 is much more attractive than $500+ :rolleyes:
Isn’t the vex explorer tread kit the same thing as the vex tread kit?
If they are it seems pointless to buy the vex explorer kit when for $5 more you can get the vex tread kit which has more links and more pieces to work with.
Just saying, i think the vex tread kit is a better value over the vex explorer kit.
Turns out, I already have a digital force gauge: My postal scale!
It is rated up to 75lbs/34kg, and has 0.1oz/0.5g resolution. I’ve verified that it is based on a load cell design, which means it will work in any orientation, and can be used for measuring either tension or compression forces. Basically, it is identical in function to the $500+ gauges I was looking at, except for the RS232 output.
The scale seems to update its display 5 to 10 times per second, so I can just videotape the experiment with the scale display in view and have an excellent way to capture the data.
I’m going to perform three basic tests:
[LIST=1]
*]Pushing/pulling strength on a level surface
*]Traction on a sloped surface (maximum ascent angle)
*]Steering efficiency (time to spin 360 degrees in place)
[/LIST]
So far, I’ve settled on the following surfaces:
[LIST=1]
*]Raw Plywood
*]Painted Plywood
*]Pergo style hardwood floor
*]Low-pile carpet
*]Foam Tiles (Vex field tiles)
*]Raw Concrete
*]Grass (push/pull and steering test only, no ramp)
[/LIST]
I am going to test all the Vex and Vexplorer wheels and tracks in a variety of configurations, which is getting to be a long list, so I won’t go into it here.
I plan to start preparing the surface panels this weekend and hope to begin collecting data next weekend. I’ll let everybody know how it’s going.
First, I’m going to set the scale on it’s back with the weighing platform facing the robot. The robot can simply line-drive into the scale and I’ll see how much pushing force is generated.
Second, the weighing platform is attached to the load cell with two bolts. I can remove them and replace one with a screw-eye. I can then attach a cable to the screw eye and let the robot pull it.
I would expect to get similar results from both tests; if not, I’ll have to stop and figure out why.
When I think of doing a traction measurement, I think of the tractor pull concept.
If you remember or lookup the physics of friction, F=uMg = u (weight).
Remember to normalize the weight of different robots to get more meaningful results.
‘u’ is the coefficient of friction, and it is greater for static (not sliding) vs dynamic (sliding friction).
I’ve thought of using a robot to string pulley at the edge of the table to lift a chain from a coil on the floor. When the robot slides backwards, you can measure the length of chain it lifted.
I’m looking forward to your results.
So, I’ve got just about everything ready:
[LIST]
*]I’ve got most of the surface samples ready
*]I’ve validated that the data collection system works and provides repeatable results.
*]I’ve compiled my (rather daunting) list of surface/wheel experiments to run
*]I’ve got a very reliable and basic robot chassis (small square chassis with 4 motors direct-driving the axles.
[/LIST]
I’m still waiting for the foam field tiles to be delivered, and I’m going to write a simple program to automate the tests a bit.
Here is a quick graph to show you guys what I’m up to. This represents an 8 sec run of a standard VEXplorer + Wrist Kit + 7.2V NiCd pack (combined weight = 1.66kg).
So, the wheel configuration is 6-wheel drive using VEXplorer wheels, driven by 2 VEXplorer motors. The motors had more power than the wheels had traction (the wheels slipped rather than the motors stalling).
The surface material is a Pergo-like laminate (synthetic hardwood floor with a “Satin Gloss” finish).
The Y axis units are grams on my postal scale, which would be Milli-Newtons as a force measurement. The peak reading was 1990g, or 1.99N
The rough plateau around 1.9N is where the motors are on. The plateau around 0.8N at the end was with the motors off. The VEXplorer was pressing hard enough against the scales that it continued to exert this much pressure. It returned to zero after I backed it away from the scale.