Accuracy of gyroscopes and accelerometers

I have heard that gyroscopes and accelerometers are a bit inaccurate, so are they accurate enough to stick something into the hanging pole during autonomous after running around the field?

1 Like

I haven’t had enough experience with them to answer your question but I remember lots of threads a year ago that taught how to make sensors more accuracy… basically you would use as many sensors as possible (encoders, gyroscope, accelerometer, ultrasonic, etc.) and multiply how far each sensor says you have gone in each direction (and turning) by a fraction that all the fractions add in each direction total of one. You multiply your most accurate sensors by a bigger fraction so the final distance is more effected by your most accurate sensor…

You may want to look up Kalman filter.

Sure that would be the perfect filter. For accel and gyro there is a simple filter called a complementary filter.

Gyro can definitely have some use. Judging by the question it sounds like those would be the only sensors and that is almost certainly not enough. I would suggest drive encoders over the accel.

Well, not sure about your drive base, but for our X drive, simply slamming into the corner aligns us perfectly. The corners of the drive are at 45 degrees, which makes it ridiculously easy to align. Sensors are great, but IMO sometimes a physical solution can work just as well, if not better :slight_smile:

The only problem is the robot doesn’t know if it just slammed into a star which is in the corner.

Oh right, I forgot about that :stuck_out_tongue: We’re only planning on hanging during driver, but in auton, that will probably be something to worry about too. Assuming you started at the close tile, it should be pretty easy to dump that one star out of the way before going for the hang (our old auton launched the preloaded star and the one in the corner in ~4 seconds).

But our robot will be running around the field before hanging.

Tracking wheels are unpowered wheels that are connected to encoders. It ends up with something similar to what VW did with their front wheels, but used for different purposes.
Tracking wheels will still be affected by any movement while not contacting the ground. They just don’t have issues with pushing battles or other wheel slip-inducing situations.

A “reset north” button to reset the gyro to 0 degrees can help if you are using the gyro actively during driver control. This is helpful if you have a field-centric holonomic drive. (basically a rotational translation for the drive to be with the gyro vs the facing of the robot)

Would it be legal for driver control, have a little period where the robot will autonomously detach from the pole then go to its default tile? Because if it is legal, you could have the robot get down, face the starting position, then the robot will reset the gyroscope relevant to where it started off to remove the issue of gyro drift from/by lifting.

And For the tracking wheels, WingusDingus has added an odometry wheel to their robot, as shown in this video at 0:46

I think it is legal.

Yeah, it should be legal. You can have autonomous functions in driver control, but it would be easier to just drive it to the tile then press a button.

True, true… Maybe the robot doesn’t have to drive at all to reset its sensors, just make the robot go down, then set the sensors to that same location the robot was down at?