Incredibly

informal
overview of
control theory
And how to control a FIRST robot
Anatomy of a
Robot
Electromechanical parts of Interest
Stepper Motor

• Moves forward one “Step”

for each pulse of
electricity
• “Open Loop Control”
Servo Motor

• Used in “Closed loop

control”
• Continuous
• Moves forward by
alternating the phased
sign waves
Motor Controller– REV Robotics, Spark MAX

Input: Control signal (1,-1)

Output: Motor health,
• A motor controller is a dumb Temperature, Encoder
computer good at three things,
• Reading encoder signals Output: A, B, C
Phases
• Making sine waves
• And reading a control signal

Input: Power
(Voltage)
Encoders – usually automatically
attached to Servos
• Two types
• Absolute
• Not used often, especially within FRC
• Relative
• Most common in FRC, What is
installed in our ‘23 robot
• You use “magnetic encoders”—Less
good than optical
• Measures distance by counting
“Pulses” can identify directions
• FIRST calls this “Odometry”
• “Route measuring”
What do encoders look like for our purposes
• Top- Cancoder
• CTRE
• Communicates using CAN bus
• Bottom Right- SRX Mag Encoder
• CTRE
• Communicates via pulse width
modulation PWM
• Bottom Left –Through bore Encoder
• Rev Robotics
• PWM
Controller–
Robo RIO
• The “Big Brain” of the robot
• Handles all input and output via
• Digital inputs/outputs
• Pulse width modulated inputs/outputs
• Analog inputs
• Relays
 Encoder
Spark Max Motor
Robo RIO
Controller
Servo motor
Battery
“But the Kraken/Falcon/Sparks/etc. don’t
have Encoders/Controllers”
What is control
theory?
Introduction to classical control
principles
Control Theory
• The field of controlling dynamic
systems
• Identify a goal, and use electrical,
mechanical, and computer systems
to reach said goal
• Examples
• Cruise Control
• Thermometer
• Heat seeking missiles
• Segways ( invented by the founder
of FRC)
Terms to know
What is a system? Its whatever. We don’t care
• Anything that takes a signal in,
modifies it, and produces a
measurable output
• Examples include
• Cars
• Stoves
• Doors System/Plant

Input/ control signal

output
Open Loop Control: No feedback
• Example case: Wind up car
1x rotation
• Input: wind up torque
• System/plant: car
• Output: distance
• We can calculate the
relationship between input-
1.25 feet
torque and output- distance
• E.g. 1 turn = 1.25 feet
Open Loop control limitations
• Wind up the car 3
times, how far will it
go? 3x rotation

• Can it handle the hill?

• Will it stay on the hill?
• What if the spring on 2 feet
the winding is non-
linear?
Closed loop control: an example
• Regular Car
• Input: Gas/Accelerator
• System: Car
• Output: Distance
• Controller: Person
• Feedback: Eyes
• The controller knows where to go, 2 feet
and when they’ve reached there.
• We tell the controller where to go,
and the controller applies as much
gas as required
 Control Theory is represented in Block
diagrams
Winding Distance
Windup Car

Wanted Position
Error Pressed Gas Pedal
(brainwaves)
- (wanted position-actual position)
Driver Real car
Distance

Brain waves
Eyes
Additional benefits of a controller
• They make unstable things
stable.
• Responsive to environments
• Increase performance.
• How do you make my ‘89 ford
Tempo function like a
Maserati?
• Essentially like replacing the
little old lady driver with a
racecar driver
PID controller
How do I make a controller?
What is a PID
Controller
WPILIB, CTRE, Rev
robotics already did
the math
What is proportional control, intuitively*
• Imagine our goal is to move
blocks from point A to Point B
• The size of the block is
representative of the error of
the system
• If block is big, we send big guy
• If block is small, we send small
guy
• *this is obviously a simplification
What is Integral control, intuitively
• Sometimes P won’t work (the
block gets stuck on a rock for
example)
• By measuring the history of the
blocks movement
measures all of the missed goal
posts and sends Integral terms
to help
• Because it’s the summation of all
past actions, the small Integral
gains “I” can add up fast
Integral Windup, What is it?
• If the controller can’t reach its
goal, the error will add up and
the integral term will get bigger
• Something has to give, the
obstruction, or the system
• If it’s the obstruction, the system
will leap forward
• If it’s the system the system
(motor, frame, gear) will break
What is a Derivative Control, Intuitively?
( )
• looks forward to where he
thinks the system will be
• He predicts the systems movement
by how fast the system accelerates Slow Down!
We are
• He sends the derivative terms to almost there
address future changes
• Can change size very rapidly
• Can be very big or very small
• Derivatives are “noisy” too much D
will cause “Chatter”
How to choose the PID constants
1. Set I and D to zero
2. Start out with a small P value
3. Give the system a reference point
4. Double the P value until the system
starts to oscillate/go unstable
5. Halve the P value and slowly creep
up until you oscillate stably
6. Slowly bring up the I value until the
oscillations disappear
7. Increase D to make the response
faster
Feed
Forward
control and
Disturbances
But Mr. Luke, I need it to go
faster/better/more accurate
 Mechanism models (According to WPILIB)
• Motor
• kS

• kV
• Voltage needed to break friction/operate at all Motor
• Voltage needed to maintain speed
• kA
• Voltage needed to change speed
• Elevator/Arm Elevator
• kG
• Voltage needed to overcome Gravity
• Other Terms
• d,𝑑̇ ,𝑑̈

• V
• distance/velocity/acceleration based on number of dots
Arm
• Output voltage
• Sgn
• Sign of the equation (positive or negative
• Θ
• angle
What tools do we have
• WPILIB has a SYSID library
• It doesn’t work great
• 2928 has a walkthrough which
may work better?
• CTRE’s Phoenix Tuner?
• Research more options
• Or build our own!
• https://2928-frc-programmer-
• Last years solution training.readthedocs.io/en/lates
• Guess and check! t/Romi/Control/romiSystemId/