Trouble with Tank Drive distances (inPerTick isn't linear?)

[a-f-waterhouse]

Hi! So I posed this question on the FTC discord and got told to open an issue here :)

I am using roadrunner for a Tank Drive, which has 2 drive motors, and I'm using the encoders in those for my localization.
My theoretical value for InPerTick is 0.07928 (from calculation based on motor + wheel diameter etc.), and this is close to some measured values of it, however my issue is that different distances result in different values form InPerTick (within about a +- 0.01 range).

In practice, while using the theoretical value of 0.07928, this means that the robot can travel most distances accurately, but not small (<10ish inches) or large (>30ish inches) distances [the error becomes more apparent for increasingly smaller/larger distances]. Based on the field view on FTC dashboard, the robot doesn't register this error as error - it thinks it has gone the correct distance when it is in fact much more/less.

My other values from the tuning process are:
kA: 0.00001
kS: 3.998110439966781
kV: 0.00481769727090664
trackWidthTicks: 126
ramseteBBar: 5
ramseteZeta: 0.7

Only things I can think of worth noting from those is that ramseteBBar is quite high (at lower values it turns far too aggressively) and I've also limited the maxWheelVel to 20, but I am not sure whether either of those are linked at all :]

[rbrott]

this is close to some measured values of it, however my issue is that different distances result in different values form InPerTick (within about a +- 0.01 range).

How are you conducting the measurements? Are you pushing the robot in a manner that ensures the wheels do not slip? ~10% (0.01 / ~0.08) error from inPerTick alone is pretty poor.

this means that the robot can travel most distances accurately, but not small (<10ish inches) or large (>30ish inches) distances [the error becomes more apparent for increasingly smaller/larger distances].

Are you sure this is attributable to inPerTick? Are you saying that the localization is inaccurate (reported position in dash/telemetry doesn't match distance traveled IRL) or rather that the robot doesn't achieve the desired distances?

Only things I can think of worth noting from those is that ramseteBBar is quite high (at lower values it turns far too aggressively) and I've also limited the maxWheelVel to 20, but I am not sure whether either of those are linked at all :]

These other parameters shouldn't matter so long as they aren't causing the wheels to slip.

[a-f-waterhouse]

How are you conducting the measurements? Are you pushing the robot in a manner that ensures the wheels do not slip? ~10% (0.01 / ~0.08) error from inPerTick alone is pretty poor.

using the tiles to make sure its in a straight line, then using a tape measure to get the distance. the wheels might be slipping slightly, but not any noticeable amount, and repeating measurements for the same distance gets very consistent results.

Are you sure this is attributable to inPerTick? Are you saying that the localization is inaccurate (reported position in dash/telemetry doesn't match distance traveled IRL) or rather that the robot doesn't achieve the desired distances?

I'm not completely confident this is the issue, its just my closest guess :) and yes, the localization is inaccurate - it reaches the position it intends to, but this just doesn't correspond with the intended distance in real life.