Breaking News
Blog

# Do Arduinos Dream of Electric Sheep?

Blog
10/17/2013 04:46 PM EDT

NO RATINGS
3 saves
Page 1 / 2 Next >
Page 1 / 2   >   >>
User Rank
Author
Ooooh...shiny!!
10/17/2013 5:04:54 PM
NO RATINGS
I'm impressed with those wheels, Max.  There are some clever guys around.....

I assume that to rotate you drive ALL motors in the same direction, and to go forward  you drive two motors only, in opposite senses (as in clockwise / anticlockwise)?

Will you be implementing algorithms that decide which motors to use for straight line motion based on the current position of the robot, or will you always use one orientation for "forward" and always rotate the robot so that that is the way it is facing?  Hope this is clear....  For example if your robot is facing with wheel A at north, B at Southeast and C at southwest, and you want to go east, you could just rotate anticlockwise about 30 degrees and then drive with wheels A and B.  But if your nominal "forward" direction is between wheels A and C, you'd have to rotate a lot more to get your nominal "forward" direction pointing east - 150 degrees in fact. But using minimum rotation would take a lot more "smarts" - though I'm sure you will be able to impart that to your robot?

User Rank
Author
Re: Ooooh...shiny!!
10/17/2013 5:55:24 PM
NO RATINGS
Interesting wheel-fu.  It reminds me of Stephen Leacock's wonderful sentence:
He flung himself from the room, flung himself upon his horse and rode madly off in all directions.

User Rank
Author
What a Mechanics toy!!
10/17/2013 5:56:15 PM
NO RATINGS
I really like the platform you have finally chosen for your robot platform. I think this is a Mechanics toy with endless possibilities.

And no, I'm not writting Mechanics with capital "M" because a typo. When looking your diagrams in the second page for illustrating some possible movements for the three wheeled "flying saucer" it seems clear to me that you are going to struggle with force vectors addition and drag ressistance in order to describe your robot dynamics.

I think that some very usefull MEMS shields for the three wheeled robot should be a 2-3 axis accelerometer and, most important, a compass module -- that's the only way I can imagine for calibrating the orientation of a robot that is going to pass most of its time turning around its own symmetry axis.

Now, let me be the first of making you this question: are you maybe thinking on building a Roomba-like cleaner for The Pleasure Dome?? LOL

User Rank
Author
Re: What a Mechanics toy!!
10/17/2013 6:18:14 PM
NO RATINGS
@Garcia - "are you maybe thinking on building a Roomba-like cleaner....."great idea Javi, with only 3 wheels, that will use the 4th motor driver on the shield!

User Rank
Author
Re: Ooooh...shiny!!
10/18/2013 9:27:25 AM
NO RATINGS
@David: I'm impressed with those wheels, Max.  There are some clever guys around.....

I agree -- I only wish I was one of them :-)

User Rank
Author
Re: Ooooh...shiny!!
10/18/2013 9:37:16 AM
NO RATINGS
@David: I assume that to rotate you drive ALL motors in the same direction, and to go forward  you drive two motors only, in opposite senses (as in clockwise / anticlockwise)?

As you say, to rotate on the spot you would drive all of the motors in the same direction.

When it comes to moving "forward" -- the depends on what you mean by "forward." Let's first assume that the robot doesn't have anything we would call a "front" (i.e., no particular sensor configuration that's special). If we decided to move it in a direction that was directly along the axis from the center through one of it's wheels, then -- as you say -- we would rotate the other two wheels (one clockwise and the other anticlockwise -- of course this may be confusing to people until they think about it a bit).

But now suppose we want to move the robot in a direction that was not along the axis from the center through one of it's wheels -- let's say it was 15 degrees different from the original case above -- in this case we might rotate our original two wheels exactly the same as before while rotating the thirs wheel a small amount -- or maybe in addition to rotating the third wheel a small amount we need to increase the rotational speed of one of our original wheels and/or decrease the rotational speed of the other ... I will have to experiment with this and report back...

User Rank
Author
Re: Ooooh...shiny!!
10/18/2013 9:45:32 AM
NO RATINGS
@David: Will you be implementing algorithms that decide which motors to use for straight line motion based on the current position of the robot, or will you always use one orientation for "forward" and always rotate the robot so that that is the way it is facing?

I think I will start off by saying that when I put the robot on the floor, it assumes it's pointing at "north" or 0 degrees, and that this curresponds to an axis that passes from its center directly through one of its three wheels (I've yet to decide whenther the 0 degrees is the side with the wheel or th eside oposite from the wheel. So all of my original tests will be along the lines of "how do I rotate it 'x' degrees" and how to I go in a straight line in a direction of 'y' degrees for a distance of 'z' centemeters.

However, on the basis that i am planning on eventually adding a Pixy Machine Vision sensor -- I will eventually say that my 0 degrees position on the robot is the "front" -- so not I want to turn it in the direction I want it to go. Of course, as I discuss in thsi blog, I can perform rotation and translation at the same time -- it's all just a matter of varying the three motor speeds and directions.

At some stage I want the robot to fully understand it's orientation with regard to the outside world -- let's say it has a magnometer so it can tell magnetic north. So now it understands 0 degrees / front (in the content of its own body) versus 0 degrees as magnetic north.

But I fear we are getting ahead of my furture experiments and blogs...

User Rank
Author
Re: Ooooh...shiny!!
10/18/2013 9:46:15 AM
NO RATINGS
@betajet: He flung himself from the room, flung himself upon his horse and rode madly off in all directions.

I love thsi saying -- thanks for sharing

User Rank
Author
Re: What a Mechanics toy!!
10/18/2013 9:49:32 AM
NO RATINGS
@Garcia: ...it seems clear to me that you are going to struggle with force vectors addition and drag ressistance in order to describe your robot dynamics.

I think that some very usefull MEMS shields for the three wheeled robot should be a 2-3 axis accelerometer and, most important, a compass module -- that's the only way I can imagine for calibrating the orientation of a robot that is going to pass most of its time turning around its own symmetry axis.

I agree -- the robit needs some way of knowing "where it is" and "what it's done" and to compare this to "where it was" and "what it was trying to do" -- not the least that I'm pondering the idea of using Genetic Algorithms to "train" it ... and they need  away of measuring how well they did.

User Rank
Author
Genetic Algorithm
10/18/2013 10:37:08 AM
NO RATINGS
Max-

How about using an algorithm that 'learns' how to move in a particular direction? The robot starts out with only some simple functions for wheel control and builds up functions for specific directions and speeds after 'trying' out some combinations. 'Winning' combinations become the next 'best' function for moving that way. Let the robot do the work of coding by programming itself...

Page 1 / 2   >   >>