Society of Robots - Robot Forum

Mechanics and Construction => Mechanics and Construction => Topic started by: kunal5959 on November 23, 2011, 10:40:26 AM

Title: Regarding Tolerance information about the Mecanum wheels
Post by: kunal5959 on November 23, 2011, 10:40:26 AM
I have  to make a 4WD 100mm Mecanum wheel robot  as my internship project . The requirements of company need Tolerance information about the Mecanum wheels which i will use  ,but while searching for mecanum wheels none of the sellers had any information about tolerances involved in their datasheet..i need this information  because we want our robot to carry a load <10 kg with great precision (in mm range) and most important issue is roller and wheel tolerances for precise and stable motion . I would like to know what kind of tolerances can be expected in a mecanum wheel. Also  what is the best possible way to remove these tolerances.

a)  For my project the main requirement is,for eg..if i program  my robot to move with constant speed  till 10.4 m and stop exactly at 10.4m will it stop or will i get an error due to roller ,bearing or other mecahnical tolerances.If yes ,does using an encoder of high resolution help us to minimize tolerance error??

b)  Also if i use a bigger wheel or a smaller diameter wheel from the same manufacturer is there a chance that bigger diameter wheel will have comparatively more tolerance error and smaller diameter wheel have less tolerance error than the 100mm wheel.???? :)
Title: Re: Regarding Tolerance information about the Mecanum wheels
Post by: newInRobotics on November 23, 2011, 11:35:24 AM
I would like to know what kind of tolerances can be expected in a mecanum wheel. Also  what is the best possible way to remove these tolerances.
You cannot remove tolerance as it is only a name to call allowable error. Although, You can minimise error, or compensate for error.

a)  For my project the main requirement is,for eg..if i program  my robot to move with constant speed  till 10.4 m and stop exactly at 10.4m will it stop or will i get an error due to roller ,bearing or other mecahnical tolerances.If yes ,does using an encoder of high resolution help us to minimize tolerance error??
The higher the resolution the higher the precision, as simple as that. No matter what resolution encoder You use, error will always be there. So, if You want Your robot to stop at 10.4m mark, it might stop at 10.45m, or 10.35m, or 10.46m mark, so it is for You to decide what tolerance (amount of error) is acceptable; remember - You will never get 100% precision, that does not exist.

b)  Also if i use a bigger wheel or a smaller diameter wheel from the same manufacturer is there a chance that bigger diameter wheel will have comparatively more tolerance error and smaller diameter wheel have less tolerance error than the 100mm wheel.Huh? Smiley
Normally, bigger things have bigger tolerances. Imagine a bullet that has allowable tolerance of 0.001g (just a guess), and a tank that has tolerance of maybe 10kg, You get my point I hope. Thee won't be that much difference between wheels of 10cm radius and 11cm radius. What You should do is contact manufacturer of wheels You want to buy and ask what tolerances they have for the products they make (or a specific product You are after).
Title: Re: Regarding Tolerance information about the Mecanum wheels
Post by: Soeren on November 23, 2011, 05:54:38 PM
Hi,

The higher the resolution the higher the precision, as simple as that.
Sorry, but I do have to protest (very rigorously) about that statement!
Resolution and precision is two different things and while they often follow each other, that's not a given thing.
You can have an extremely high resolution and a lousy precision (and v.v.).
To take a recent example, imagine using a common threaded rod (with a pitch of 1mm), using a 100:1 gearbox and a 10 segment encoder on the driving axle of the gearbox - resolution=1µm, precision=depends on how the rod was threaded, but nowhere near the resolution.
Using a 100 segment encoder on the same setup, your resolution would change to 100nm, but would the precision change any?
The reverse could be true as well. Imagine the same but with only 1 pulse per revolution, or one in 10 revs for that matter - a lousy resolution compared to the precision.


kunal5959 <-
The real question is, how can anyone ask about tolerances, ask about 1mm pecision and in the same post say 1.4m?

What is 1.4m?
As I read it, anything between 1.350m and 1.449m is within the given precision, as it is specified with a single digit.

If a single mm +/- is the Grail, the number should be 1.400m (or more correct in the mech business, 1,400mm).

Just as a sanity check, 1mm in 1.4m is a precision of 0.07% and you won't get that, whether you use Mechanums or eg. rigid aluminum wheels.

When you make specifications for a project, always make plenty of sanity checks along the way - wishful thinking is for dreamers, engineering (however mundane) needs hard facts to be successful :)


Why do you think a precision of 0.07% (or 700 ppm) is important in this mechanical setup?

If it really is (which I somehow doubt), either throw in the towel or start looking for other means of achieving this by compensation.
Like, if you need it to load something, you could use optical feedback while moving the last bit forward and/or back until you're in the precise spot.
Title: Re: Regarding Tolerance information about the Mecanum wheels
Post by: newInRobotics on November 24, 2011, 01:20:04 AM
Sorry, but I do have to protest (very rigorously) about that statement!
Resolution and precision is two different things and while they often follow each other, that's not a given thing.
You can have an extremely high resolution and a lousy precision (and v.v.).
To take a recent example, imagine using a common threaded rod (with a pitch of 1mm), using a 100:1 gearbox and a 10 segment encoder on the driving axle of the gearbox - resolution=1µm, precision=depends on how the rod was threaded, but nowhere near the resolution.
Using a 100 segment encoder on the same setup, your resolution would change to 100nm, but would the precision change any?
The reverse could be true as well. Imagine the same but with only 1 pulse per revolution, or one in 10 revs for that matter - a lousy resolution compared to the precision.
I did not give it a good thought  ;D At the time I was thinking along lines of shaft encoder and PWM driven motor speed control.