03-20-2025, 08:44 AM
(This post was last modified: 03-24-2025, 12:40 PM by noitarud.
Edit Reason: Terrain, speeds
)
Here is my idea of how tracking is possible with just one 3-axis accelerometer. The motion of the accelerometer on the buggy has to be confined to certain freedoms. One axis is dedicated to thrust perception, the second is for centrifuge perception, the third for gravity. The accelerometer must be positioned at the centre of rotation (pivot), usually the fixed rear axle, this prevents centrifuge anomalies from buggy circling (i presume some lateral misposition of a'mtr will need compensation for radius and tilt progression (as opposed to absolute tilt)). If gravity cancels the centrifuge/thrust during tilt, the direction/descent has to be inferred from preceding events (though tilting is not ideal and lumpy terrain will complicate, the uncertainty is known due to z-axis anomaly and is reduced if the RPM is known). Loss of pivot's traction (speed; or bump) will lose centrifuge for unperceived lateral thrust (may be inferrable to a degree? If speed, will have to assume any of the front wheels has become the pivot to their friction's degree). Sudden changes of z-axis can map the terrain, will have to tolerate bounce when crossing seams in the flooring.
applications: Explore the size of rooms, do passes of furnishings to map them…
You need speed for the centrifuge reading to begin to work (cannot drive wheels in opposite directions), and accurately known speed to get a useful result. To correct cumulated speed flaws, try stopping (level) or measuring the seams in the floor. But with terrain, can you also compute pitch and roll speeds and know how their centrifugal forces and gravity adds to the centrifuge and thrust measurements as well? which shows the limit, you can only do one: pitch or roll. Placing the sensor off-centre will work at zero speed but there will be a (chiral) radius where the sensitivity will die out so you will need to estimate how close you are to reaching it, consider doing a rehearsed manouver to cross the threshhold, or maybe using a faulty servo that has the pot turned so the worn section is at that radius and it won't dwell there.
applications: Explore the size of rooms, do passes of furnishings to map them…
You need speed for the centrifuge reading to begin to work (cannot drive wheels in opposite directions), and accurately known speed to get a useful result. To correct cumulated speed flaws, try stopping (level) or measuring the seams in the floor. But with terrain, can you also compute pitch and roll speeds and know how their centrifugal forces and gravity adds to the centrifuge and thrust measurements as well? which shows the limit, you can only do one: pitch or roll. Placing the sensor off-centre will work at zero speed but there will be a (chiral) radius where the sensitivity will die out so you will need to estimate how close you are to reaching it, consider doing a rehearsed manouver to cross the threshhold, or maybe using a faulty servo that has the pot turned so the worn section is at that radius and it won't dwell there.
). It could get a bit better with some additional calibration, I suppose…
)