compass for navigation

[forwheeler]

I wanted to know if anyone has experience with a magnetometer or digital
compass. I assume they are the same sensor but with two names.
I am building a robot lawn mower and wanted to know if a compass would allow the robot to cross the lawn in a pattern using a north south or east west heading.

I have never used this sensor but I read they need to be calibrated and are sensitive to tilting.

I would like to hear your opinion on a part that doesn't need calibration and is accurate enough for this purpose.

[Cbenson]

Yes, a tilt-compensated magnetic compass would allow your robotic lawn mower to travel in a specific direction. You can get away with using a non-compensated compass only if your lawn is relatively level. If you have a few slopes, tilt compensation is very beneficial. If the tilt compensation is accelerometer-based, no calibration is necessary, since it knows the angle. If it's gyroscope-based, then you may need calibration.

[forwheeler]

Yes, a tilt-compensated magnetic compass would allow your robotic lawn mower to travel in a specific direction. You can get away with using a non-compensated compass only if your lawn is relatively level. If you have a few slopes, tilt compensation is very beneficial. If the tilt compensation is accelerometer-based, no calibration is necessary, since it knows the angle. If it's gyroscope-based, then you may need calibration.

Thanks for the information. According to my math I would need 1 degree accuracy otherwise it could put the mower several feet off the correct destination at the other end of the yard. It doesn't look like these devices offer that level of accuracy.
I need some other sensor to make sure it mows in straight rows. Since this is outdoors I don't think any IR type sensor would help.
The commercial mowers used a buried cable to keep it in the borders of the yard but that seems inefficient to me. Do you have any other ideas?

[Cbenson]

Some of the units do offer 0.5 to 1 degree accuracy such as RB-Dev-45. There are a variety of additional sensors you might consider such as encoders and GPS, but shouldn't there be some overlap in the mowing pattern?

[forwheeler]

I must have missed seeing that compass. It is however not listed as .5 degree accuracy but .5% which is 1.8 degrees but yes this seems like a very good option.
Yes it needs to overlap rows so, assuming a 12" cut, my plan was to mow to the other side of the yard, turn 90 degrees, drive for say 10", turn 90 degrees, then drive to the other side of the yard. I know that is ideal but it is a start.
I have no experience with IR but a coded IR light at the end of the rows would allow it to check its position once it reached the end. I don't think this will work outdoors though.
If you have more ideas please let me know.

[Cbenson]

A combination of compass, encoders and GPS should work relatively well, and be self-contained on the robot itself.

[forwheeler]

I figured a GPS module would only be accurate enough to determine if it was still in the yard but not for positioning in rows.

[Cbenson]

It depends on how large your terrain is - perhaps I was imagining something a bit too large
The other way is to try a different pattern instead of rows - perhaps a spiral?
http://www.popularmechanics.com/home...-lawn-patterns

You could even consider an approach similar to this, but use the sensors when the robot encounters an obstacle:
https://www.youtube.com/watch?v=dWW--UY2X20

[forwheeler]

The yard is maybe 70 ft X 30 ft so medium size I guess.
A spiral may make sense because you could possibly use odometry for that.
I use the banebot motors and gearboxes so I don't see a good way to use the typical encoders. I am thinking about using a HAL effect sensor for that and place magnets on the wheel hub.