Planning

[PeterMitrano]

Goals for this year

• [ ] time optimal ArcTurn and ForwardN as new behaviors
• [ ] <43 second solve

Things to do

• [x] Profile the controls code to determine how long it takes to read sensors, compute position, and actuate motors (consider PWM frequency of motors)
• [ ] Try running existing code at 500Hz, 1kHz
• [ ] more simulation?

Things we know/want

• [x] We know our encoders don't have enough CPR for low speeds
• [ ] Faster (>10kHz?) PWM to our motors otherwise our control loop is faster than our motor PWM
• [x] Our sensors apparently only actually update at 60Hz?!?

Design decisions to make

• [x] what MCU?
• [x] what encoders?
• [x] what wall sensors?

[PeterMitrano]

Things we can do to make it faster:

• use pre-computed (during init after calibration) lookup table instead of for loop and brief lookup table to make sensor computation faster
• do background analog reading/writing, and using an adc library that is faster than analogRead

Things to do:

• measure how long it takes to do encoder counting and reading (compare to ~15 microseconds for SPI @ MHz)
• use pre-computed velocity profiles with varying rates for PID control (low level) and velocity commands (high level setpoints) to determine if we need a faster PID rater, or control rate, or both, or neither

[PeterMitrano]

• As per our analysis, we need to acheive 650Hz control. This implies 650Hz sensing.
  • So we need sensors that can read distance in less than 0.0015s, which is 1.5ms or 1,500us.
• Our current compute seems more than capable of achieving this control loop rate.

[PeterMitrano]

info has been moved to wiki and TODO issues