How precise is the S2's drawing?

Oldbitcollector (Jeff)

Ok, so I've seen some of the video showing the S2 drawing words.. Pretty cool stuff.

I'm curious.

Could the S2 be used to draw cut marks on a sheet of material and get it reasonably precise?
I can see the S2 being used as a tool for making other projects.

OBC

Whit

I think it would be worth a try. After you made a drawing - you could check important specifics.

With calibration - it is pretty accurate. Also - there is amazing repeatability between different S2s using the same code. For example, I sent some code to Phil Pilgrim and his S2 responed the same as mine. It let him help me correct some things that were not working in my code as regards calibration.

I say give it a try!

W9GFO

I'd like to know more about this calibration you speak of. I have found that there is significant drift after drawing a couple letters. In my "Saints" video with the scroll underneath the word, the coordinates of the last bit of drawing had to be adjusted by 30 units to prevent the "S" from being drawn over.

Rich H

erco

Rich: I was blown away by your video. INCREDIBLE. How did you write that program?

The wheels on S2s have a fair amount of play in them. Wiggle them sideways and you can see. That will add some error that will accumulate in a complex series of maneuvers such as yours. I took the same slop out of my S1 by carefully filling the gaps between the gearmotor shaft and wheel hub adapter with gap-filling superglue. It takes patience, you have to keep testing it to make sure it runs true as you sneak up on the final position. That's on my to do list for my S2.

Phil Pilgrim (PhiPi)

There are two variables internal to the S2 object that control drawing precision: Wheel_space and Full_circle. The first relates to the actual spacing between the S2's wheels; the second, to the actual diameter of the wheels. These are initialized either to values stored in EEPROM or, if no valid values have been saved, to default values specified as constants. You can adjust both of the calibration values using the set_wheel_calibration method, as shown here with a program provided by Whit:



Once you've found calibration values that work for a particular bot on a particular surface, you can save them using the write_wheel_calibration method.

-Phil

Humanoido

Obviously the wheels will end up with some error compared to the theoretical program positions. I was reading this thread about the repeatability and have an idea. If you put four steady and stationary walls around the robot within the distance range of some Parallax Pings))) it could be used for accurate reckoning.

W9GFO

erco wrote: »

Rich: I was blown away by your video. INCREDIBLE. How did you write that program?

The wheels on S2s have a fair amount of play in them. Wiggle them sideways and you can see. That will add some error that will accumulate in a complex series of maneuvers such as yours.

Thanks! One of the students in the robotics club provided me with a spreadsheet of all the coordinates and radiuses. I scaled them and entered each manually. I have the feeling that fixing the play, the variable distance between the points of contact, and setting the exact Wheel_space and Full_circle values will make a remarkable improvement in accuracy.

Rich H

erco

The S2 is a great platform, we're all excited about the potential. Processor, electronics & sensor-wise, we'll all be busy for a while. I think the first limiter maybe the plastic body & wheels. IMHO, this could scale up to a very nice mid-size robot for household navigation. Gut an S2 and transfer the electronics to bigger platform with stronger motors & more rigid wheels. Even a Stingray. Oh, the possibilities!