frc971/control_loops/python/haptic_wheel.py

#!/usr/bin/python

from frc971.control_loops.python import control_loop
from frc971.control_loops.python import controls
import numpy
import sys
import copy
import scipy.interpolate
from matplotlib import pylab

import gflags
import glog

FLAGS = gflags.FLAGS

gflags.DEFINE_bool('plot', False, 'If true, plot the loop response.')
gflags.DEFINE_string('data', None, 'If defined, plot the provided CAN data')
gflags.DEFINE_bool('rerun_kf', False, 'If true, rerun the KF.  The torque in the data file will be interpreted as the commanded current.')

class SystemParams(object):
  def __init__(self, J, G, kP, kD, kCompensationTimeconstant, q_pos, q_vel,
               q_torque, current_limit):
    self.J = J
    self.G = G
    self.q_pos = q_pos
    self.q_vel = q_vel
    self.q_torque = q_torque
    self.kP = kP
    self.kD = kD
    self.kCompensationTimeconstant = kCompensationTimeconstant
    self.r_pos = 0.001
    self.current_limit = current_limit

    #[15.0, 0.25],
    #[10.0, 0.2],
    #[5.0, 0.13],
    #[3.0, 0.10],
    #[2.0, 0.08],
    #[1.0, 0.06],
    #[0.5, 0.05],
    #[0.25, 0.025],

kWheel = SystemParams(J=0.0008,
                      G=(1.25 + 0.02) / 0.35,
                      q_pos=0.001,
                      q_vel=0.20,
                      q_torque=0.005,
                      kP=7.0,
                      kD=0.0,
                      kCompensationTimeconstant=0.95,
                      current_limit=4.5)
kTrigger = SystemParams(J=0.00025,
                        G=(0.925 * 2.0 + 0.02) / 0.35,
                        q_pos=0.001,
                        q_vel=0.1,
                        q_torque=0.005,
                        kP=120.0,
                        kD=1.8,
                        kCompensationTimeconstant=0.95,
                        current_limit=3.0)

class HapticInput(control_loop.ControlLoop):
  def __init__(self, params=None, name='HapticInput'):
    # The defaults are for the steering wheel.
    super(HapticInput, self).__init__(name)
    motor = self.motor = control_loop.MN3510()

    # Moment of inertia of the wheel in kg m^2
    self.J = params.J

    # Control loop time step
    self.dt = 0.001

    # Gear ratio from the motor to the input.
    self.G = params.G

    self.A_continuous = numpy.matrix(numpy.zeros((2, 2)))
    self.A_continuous[1, 1] = 0
    self.A_continuous[0, 1] = 1

    self.B_continuous = numpy.matrix(numpy.zeros((2, 1)))
    self.B_continuous[1, 0] = motor.Kt * self.G / self.J

    # State feedback matrices
    # [position, angular velocity]
    self.C = numpy.matrix([[1.0, 0.0]])
    self.D = numpy.matrix([[0.0]])

    self.A, self.B = self.ContinuousToDiscrete(
        self.A_continuous, self.B_continuous, self.dt)

    self.U_max = numpy.matrix([[2.5]])
    self.U_min = numpy.matrix([[-2.5]])

    self.L = numpy.matrix([[0.0], [0.0]])
    self.K = numpy.matrix([[0.0, 0.0]])

    self.InitializeState()


class IntegralHapticInput(HapticInput):
  def __init__(self, params=None, name="IntegralHapticInput"):
    super(IntegralHapticInput, self).__init__(name=name, params=params)

    self.A_continuous_unaugmented = self.A_continuous
    self.B_continuous_unaugmented = self.B_continuous

    self.A_continuous = numpy.matrix(numpy.zeros((3, 3)))
    self.A_continuous[0:2, 0:2] = self.A_continuous_unaugmented
    self.A_continuous[1, 2] = (1 / self.J)

    self.B_continuous = numpy.matrix(numpy.zeros((3, 1)))
    self.B_continuous[0:2, 0] = self.B_continuous_unaugmented

    self.C_unaugmented = self.C
    self.C = numpy.matrix(numpy.zeros((1, 3)))
    self.C[0:1, 0:2] = self.C_unaugmented

    self.A, self.B = self.ContinuousToDiscrete(
        self.A_continuous, self.B_continuous, self.dt)

    self.Q = numpy.matrix([[(params.q_pos ** 2.0), 0.0, 0.0],
                           [0.0, (params.q_vel ** 2.0), 0.0],
                           [0.0, 0.0, (params.q_torque ** 2.0)]])

    self.R = numpy.matrix([[(params.r_pos ** 2.0)]])

    self.KalmanGain, self.Q_steady = controls.kalman(
        A=self.A, B=self.B, C=self.C, Q=self.Q, R=self.R)
    self.L = self.A * self.KalmanGain

    self.K_unaugmented = self.K
    self.K = numpy.matrix(numpy.zeros((1, 3)))
    self.K[0, 0:2] = self.K_unaugmented
    self.K[0, 2] = 1.0 / (self.motor.Kt / (self.motor.resistance))

    self.InitializeState()

def ReadCan(filename):
  """Reads the candump in filename and returns the 4 fields."""
  trigger = []
  trigger_velocity = []
  trigger_torque = []
  trigger_current = []
  wheel = []
  wheel_velocity = []
  wheel_torque = []
  wheel_current = []

  trigger_request_time = [0.0]
  trigger_request_current = [0.0]
  wheel_request_time = [0.0]
  wheel_request_current = [0.0]

  with open(filename, 'r') as fd:
    for line in fd:
      data = line.split()
      can_id = int(data[1], 16)
      if can_id == 0:
        data = [int(d, 16) for d in data[3:]]
        trigger.append(((data[0] + (data[1] << 8)) - 32768) / 32768.0)
        trigger_velocity.append(((data[2] + (data[3] << 8)) - 32768) / 32768.0)
        trigger_torque.append(((data[4] + (data[5] << 8)) - 32768) / 32768.0)
        trigger_current.append(((data[6] + ((data[7] & 0x3f) << 8)) - 8192) / 8192.0)
      elif can_id == 1:
        data = [int(d, 16) for d in data[3:]]
        wheel.append(((data[0] + (data[1] << 8)) - 32768) / 32768.0)
        wheel_velocity.append(((data[2] + (data[3] << 8)) - 32768) / 32768.0)
        wheel_torque.append(((data[4] + (data[5] << 8)) - 32768) / 32768.0)
        wheel_current.append(((data[6] + ((data[7] & 0x3f) << 8)) - 8192) / 8192.0)
      elif can_id == 2:
        data = [int(d, 16) for d in data[3:]]
        trigger_request_current.append(((data[4] + (data[5] << 8)) - 32768) / 32768.0)
        trigger_request_time.append(len(trigger) * 0.001)
      elif can_id == 3:
        data = [int(d, 16) for d in data[3:]]
        wheel_request_current.append(((data[4] + (data[5] << 8)) - 32768) / 32768.0)
        wheel_request_time.append(len(wheel) * 0.001)

  trigger_data_time = numpy.arange(0, len(trigger)) * 0.001
  wheel_data_time = numpy.arange(0, len(wheel)) * 0.001

  # Extend out the data in the interpolation table.
  trigger_request_time.append(trigger_data_time[-1])
  trigger_request_current.append(trigger_request_current[-1])
  wheel_request_time.append(wheel_data_time[-1])
  wheel_request_current.append(wheel_request_current[-1])

  return (trigger_data_time, wheel_data_time, trigger, wheel, trigger_velocity,
          wheel_velocity, trigger_torque, wheel_torque, trigger_current,
          wheel_current, trigger_request_time, trigger_request_current,
          wheel_request_time, wheel_request_current)

def rerun_and_plot_kf(data_time, data_radians, data_current, data_request_current,
                      params, run_correct=True):
  kf_velocity = []
  dt_velocity = []
  kf_position = []
  adjusted_position = []
  last_angle = None
  haptic_observer = IntegralHapticInput(params=params)

  # Parameter sweep J.
  num_kf = 1
  min_J = max_J = params.J

  # J = 0.0022
  #num_kf = 15
  #min_J = min_J / 2.0
  #max_J = max_J * 2.0
  initial_velocity = (data_radians[1] - data_radians[0]) * 1000.0

  def DupParamsWithJ(params, J):
    p = copy.copy(params)
    p.J = J
    return p
  haptic_observers = [IntegralHapticInput(params=DupParamsWithJ(params, j)) for j in
                      numpy.logspace(numpy.log10(min_J),
                                     numpy.log10(max_J), num=num_kf)]
  # Initialize all the KF's.
  haptic_observer.X_hat[1, 0] = initial_velocity
  haptic_observer.X_hat[0, 0] = data_radians[0]
  for observer in haptic_observers:
    observer.X_hat[1, 0] = initial_velocity
    observer.X_hat[0, 0] = data_radians[0]

  last_request_current = data_request_current[0]
  kf_torques = [[] for i in xrange(num_kf)]
  for angle, current, request_current in zip(data_radians, data_current,
                                             data_request_current):
    # Predict and correct all the parameter swept observers.
    for i, observer in enumerate(haptic_observers):
      observer.Y = numpy.matrix([[angle]])
      if run_correct:
        observer.CorrectObserver(numpy.matrix([[current]]))
      kf_torques[i].append(-observer.X_hat[2, 0])
      observer.PredictObserver(numpy.matrix([[current]]))
      observer.PredictObserver(numpy.matrix([[current]]))

    # Predict and correct the main observer.
    haptic_observer.Y = numpy.matrix([[angle]])
    if run_correct:
      haptic_observer.CorrectObserver(numpy.matrix([[current]]))
    kf_position.append(haptic_observer.X_hat[0, 0])
    adjusted_position.append(kf_position[-1] - last_request_current / params.kP)
    last_request_current = last_request_current * params.kCompensationTimeconstant + request_current * (1.0 - params.kCompensationTimeconstant)
    kf_velocity.append(haptic_observer.X_hat[1, 0])
    if last_angle is None:
      last_angle = angle
    dt_velocity.append((angle - last_angle) / 0.001)

    haptic_observer.PredictObserver(numpy.matrix([[current]]))
    last_angle = angle

  # Plot the wheel observers.
  fig, ax1 = pylab.subplots()
  ax1.plot(data_time, data_radians, '.', label='wheel')
  ax1.plot(data_time, dt_velocity, '.', label='dt_velocity')
  ax1.plot(data_time, kf_velocity, '.', label='kf_velocity')
  ax1.plot(data_time, kf_position, '.', label='kf_position')
  ax1.plot(data_time, adjusted_position, '.', label='adjusted_position')

  ax2 = ax1.twinx()
  ax2.plot(data_time, data_current, label='data_current')
  ax2.plot(data_time, data_request_current, label='request_current')

  for i, kf_torque in enumerate(kf_torques):
    ax2.plot(data_time, kf_torque,
               label='-kf_torque[%f]' % haptic_observers[i].J)
  fig.tight_layout()
  ax1.legend(loc=3)
  ax2.legend(loc=4)

def plot_input(data_time, data_radians, data_velocity, data_torque,
               data_current, params, run_correct=True):
  dt_velocity = []
  last_angle = None
  initial_velocity = (data_radians[1] - data_radians[0]) * 1000.0

  for angle in data_radians:
    if last_angle is None:
      last_angle = angle
    dt_velocity.append((angle - last_angle) / 0.001)

    last_angle = angle

  # Plot the wheel observers.
  fig, ax1 = pylab.subplots()
  ax1.plot(data_time, data_radians, '.', label='angle')
  ax1.plot(data_time, data_velocity, '-', label='velocity')
  ax1.plot(data_time, dt_velocity, '.', label='dt_velocity')

  ax2 = ax1.twinx()
  ax2.plot(data_time, data_torque, label='data_torque')
  ax2.plot(data_time, data_current, label='data_current')
  fig.tight_layout()
  ax1.legend(loc=3)
  ax2.legend(loc=4)

def main(argv):
  if FLAGS.plot:
    if FLAGS.data is None:
      haptic_wheel = HapticInput()
      haptic_wheel_controller = IntegralHapticInput()
      observer_haptic_wheel = IntegralHapticInput()
      observer_haptic_wheel.X_hat[2, 0] = 0.01

      R = numpy.matrix([[0.0], [0.0], [0.0]])

      control_loop.TestSingleIntegralAxisSquareWave(
          R, 1.0, haptic_wheel, haptic_wheel_controller, observer_haptic_wheel)
    else:
      # Read the CAN trace in.
      trigger_data_time, wheel_data_time, trigger, wheel, trigger_velocity, \
          wheel_velocity, trigger_torque, wheel_torque, trigger_current, \
          wheel_current, trigger_request_time, trigger_request_current, \
          wheel_request_time, wheel_request_current = ReadCan(FLAGS.data)

      wheel_radians = [w * numpy.pi * (338.16 / 360.0) for w in wheel]
      wheel_velocity = [w * 50.0 for w in wheel_velocity]
      wheel_torque = [w / 2.0 for w in wheel_torque]
      wheel_current = [w * 10.0 for w in wheel_current]
      wheel_request_current = [w * 2.0 for w in wheel_request_current]
      resampled_wheel_request_current = scipy.interpolate.interp1d(
          wheel_request_time, wheel_request_current, kind="zero")(wheel_data_time)

      trigger_radians = [t * numpy.pi * (45.0 / 360.0) for t in trigger]
      trigger_velocity = [t * 50.0 for t in trigger_velocity]
      trigger_torque = [t / 2.0 for t in trigger_torque]
      trigger_current = [t * 10.0 for t in trigger_current]
      trigger_request_current = [t * 2.0 for t in trigger_request_current]
      resampled_trigger_request_current = scipy.interpolate.interp1d(
          trigger_request_time, trigger_request_current, kind="zero")(trigger_data_time)

      if FLAGS.rerun_kf:
        rerun_and_plot_kf(trigger_data_time, trigger_radians, trigger_current,
                          resampled_trigger_request_current, kTrigger, run_correct=True)
        rerun_and_plot_kf(wheel_data_time, wheel_radians, wheel_current,
                          resampled_wheel_request_current, kWheel, run_correct=True)
      else:
        plot_input(trigger_data_time, trigger_radians, trigger_velocity,
                   trigger_torque, trigger_current, kTrigger)
        plot_input(wheel_data_time, wheel_radians, wheel_velocity, wheel_torque,
                   wheel_current, kWheel)
      pylab.show()

    return

  if len(argv) != 9:
    glog.fatal('Expected .h file name and .cc file name')
  else:
    namespaces = ['frc971', 'control_loops', 'drivetrain']
    for name, params, filenames in [('HapticWheel', kWheel, argv[1:5]),
                                    ('HapticTrigger', kTrigger, argv[5:9])]:
      haptic_input = HapticInput(params=params, name=name)
      loop_writer = control_loop.ControlLoopWriter(name, [haptic_input],
                                                   namespaces=namespaces,
                                                   scalar_type='float')
      loop_writer.Write(filenames[0], filenames[1])

      integral_haptic_input = IntegralHapticInput(params=params,
                                                  name='Integral' + name)
      integral_loop_writer = control_loop.ControlLoopWriter(
          'Integral' + name, [integral_haptic_input], namespaces=namespaces,
          scalar_type='float')

      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "Dt", "%f",
                                integral_haptic_input.dt))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "FreeCurrent", "%f",
                                integral_haptic_input.motor.free_current))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "StallTorque", "%f",
                                integral_haptic_input.motor.stall_torque))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "J", "%f",
                                integral_haptic_input.J))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "R", "%f",
                                integral_haptic_input.motor.resistance))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "T", "%f",
                                integral_haptic_input.motor.Kt))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "V", "%f",
                                integral_haptic_input.motor.Kv))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "P", "%f",
                                params.kP))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "D", "%f",
                                params.kD))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "G", "%f",
                                params.G))
      integral_loop_writer.AddConstant(
          control_loop.Constant("k" + name + "CurrentLimit", "%f",
                                params.current_limit))

      integral_loop_writer.Write(filenames[2], filenames[3])


if __name__ == '__main__':
  argv = FLAGS(sys.argv)
  sys.exit(main(argv))