I carried out both static and dynamic pull tests for some of the wheels available for the VEX design system on the VEX playing field tiles. More on this can be found at :
First of all, I would like to thank you for doing this. This will definitely be very useful information. It is always great to see such carefully obtained empirical data like this. Perhaps you could do a test of the mecanum wheels on your setup if you get them.
this is rather startling, i was looking forward to using the âhigh tractionâ tires. so much for that. some vex products engineers need to do some investigation.
BJC sent me this video over a week ago, and I Calculated the COF in the forward direction of each of these (rounded to two digits).
COF of wheels:
Vexplorer 0.8
4â omni 0.9
2.75â double omni 0.84
2.75â single omni 0.59
2.75â green wheel 0.59
2.75â sans tread 0.53
4â stock tread 0.6
4â sans tread 0.69
4â green tread 0.68
5â wheel 0.84
5â sans tread 0.67
What people donât seem to see is that due to the surface being foam (not a smooth hard surface), wheel that tend to dig in have more âgripâ. Thus, wheels like the 5" with tread (has spike like tread), 4" without tread (has those small square indents), omni-directional (the rollers and edges dig in), and vexplorer wheels (self explanatory) have the highest traction or âgripâ on the surface given.
If the surface was glass, I can guarantee you that the high-traction wheels or the 2.75" green wheels will perform substantially better, and that the wheels with no tread will perform worse as it is hard plastic.
interesting how the rubber makes it LESS grippyâŚ
what if you tried this with a 15 pound weight (typical competition robot)
the results may be different because the wheels would sink more into the foam
I donât think that the COF will be the same for different robot weights, as this is a foam surface, and more weight will make the wheels sink in more, which will change the COF.
Empirical evidence is great!
Can you share your methodology for the calculated COF?
For instance:
What weight did you assume for the frame?
What weight did you assume for each of the wheel combinations?
How sensitive is each CoF Calculation to
these assumed weights?
imprecision in the pull test value?
In other words, how big is the error bar around these values?
For other people repeating this type of experiment,
try picking just two of your favorite wheel combinations,
and try them with two or even three different weights.
This will help show if the CoF model holds up well or not.
Its also nice to have hear some interpretation of what the data means:
4" and 5" bare wheels are similar, similar notched plastic.
4" omniâs have long been popular for better traction than 4" original tires, and it looks like the new 2.75 double omni are not far behind.
5" original tires may show high CoF because their pointy treads dig into the foam, which is a more complicated traction model than just classic CoF*mg
For impatient people, a table summary of the second video showing the results of the dynamic head to head tests would be nice.
When I get time I will retry this experiment with a 15lb robot (Maybe I can convince the physics teacher to use this as a physics experiment�)
I assumed the weight specified in the video (7lb and a few ounces?)
I then used the forward pull force as the pull force
The actual calculation is not hard (COF is force/force, so units cancel). I just wrote a spreadsheet to calculate the weight in decimal (Based on the sum of the pounds and scaled ounces), and divided it by the vehicle weight specified. The COF does not care about surface area.
The two digits was guessed, based on a similar physics experiment last year (in Hnrs. Physics). I had 7 groups of data, and they werenât all that close to each other.
The physics teacher should be more impressed if you include error bars.
If they didnât weigh the robots with the wheels each time, then the weights are different for each setup.
Use the spreadsheet to show how much difference in CoF you get if the weight varies by the weight of 4x5" wheels, in either direction. If it is > 0.05, that shows it may be important to recalculate (or measure) the weight of each setup.
Similarly, what is the implied accuracy of the scale? 1oz? 0.1lb? how much does that affect CoF?
Similarly, what is the accuracy of the 90 degree pull angle, does pulling at 85 or 95 change CoF much? or if you do the experiment yourself, can you use fixed pulley height to control the angle more accurately?
For head-to-head contests, it might also be interesting to change weights until they tie: eg âa 5 lbs 4" wheel robot is tied in pushing with a 4 lbs 4" Omniâ
Or make a long robot with a pair of wheels (of various types) at each end,
geared at chained to a common motor, pulling with high torque in opposite directions. Wheel torque doesnât affect the friction torque, right?
Awesome feedback! The scale used in the static pull viideo was a digital fish scale from Wal-Mart and it has a max pull of 50lbs and is accurate to ounces. Each static test was repeated several times and typically there was a +/- 3 ounce variation when the wheels began to slip. The weight of the robot varies due to the different weight of the wheels and in the case of the 2.75" wheels the drive gears were removed so there would be no interaction between the gears and the floor tiles. The comment that the interaction between the wheels and the field tiles are dramically different than what the results would be on a different surface is certainly accurate, but the game surface was used because for VRC teams the tile surface would be of the most interest. The goal of the test was to have teams think about what type of wheels they are going to use and hopefully test wheels for themself.
I have a set of mecanum wheels on order an hopefully they will be in this week and there should be some more data on the simplerobotics.org page by next week.