Fix wheel velocity commands to be metric units

[ashaw596]

Notes: send bodyx and bodyY command to robot which gets converted to motor signals. Soccer program is using units, but no units associated with the robot motor controls. @ashaw596

[ashaw596]

The Gearbox ratio is 4 to 1

[ashaw596]

The wheels are 40 degrees slanted more towards front and back.

[ashaw596]

From the radio-protocol-2011.txt Physical units for 2011 robots: Times are given in control cycles (5ms/cycle). 1440_4.5 = 6480 ticks/rev Distances are given in encoder ticks. 0.026_2_pi m/rev / (1440_4.5 ticks/rev) = 0.026_2_pi/(1440_4.5) m/tick Wheel angular distance: (2_pi rad/rev) / (14404.5 ticks/rev) = pi/3240 rad/tick Robot angular distance: pi/3240 rad/tick * 0.026 / 0.0812 = pi.026/(.0812*3240) robot-rad/tick = 203/776000 rad/tick

[ashaw596]

90 between each wheel.

[ashaw596]

About 5750 Encoder Ticks per revolution

[ashaw596]

About 120 cm per 5 wheel revolutiosn moving sideways one 2011 robots.

[ashaw596]

5750 * 5 / 120 = About 240 encoder ticks per centermeter

[ashaw596]

About 240 encoder ticks per centermeter * 0.005 seconds between encoder checks. = about 1.2 encoder ticks per (centermeter/ second)

[ehuang3]

@ashaw596 Here are a few comments:

• When we drive the robots through soccer, we currently cap vel and accel at 2m/s and 0.5m/s/s respectively. In the past, we've caused mechanical damage when driving above those limits. Keep that in mind w/r/t motor commands as we change things.
• FYI, our motors have not burn out in recent years and there are several safety mechanisms in place for that, including stall detection. Most damages have been mechanical.
• The 5m/cycle control loop time is strictly enforced, there should be no jitter in that. If the period changes, then the system is no longer LTI, and this breaks every assumption of ECE2025. Thus, there is no need to introduce a config scale factor for this.
• After checking the current motor command protocol (see https://github.com/RoboJackets/robocup-software/blob/master/soccer/radio/USBRadio.cpp#L231 and https://github.com/RoboJackets/robocup-software/blob/master/firmware/robot/cpu/control.c#L194) it appears that the motors commands are already in "metric" units. They are converted from m/s to encoder ticks per cycle (and clamped to the max motor command). The 1/sqrt(2) factor is a result of our 45 deg wheels assumption.
• This means that the only remaining problem is to account for the 40 deg wheels. You should be able to do this using the dot product between wheel direction and command.

Sorry for not grasping what the issues actually were earlier. Keep up the good work!

[barulicm]

Sorry, too many buttons :(

[justbuchanan]

:trollface: