lern-tracking-system/raspberry-pi/sensors/acousticSensor.py

import time
import statistics
import math
import threading
import random
import traceback
import queue

from sensors.connection import globalArduinoSlave
import logHandler

conn = globalArduinoSlave()


class AcousticSensor:
  def __init__(self, conf, calibration_state):
    self.conf = conf
    self.queue                  = queue.Queue()
    self.calibration_state      = calibration_state
    self.field_height           = float(conf["field"]["y"])
    self.field_width            = float(conf["field"]["x"])
    self.sensor_y_offset        = float(conf["ac_sensor"]["y_offset"])
    self.left_sensor_x_offset   = float(conf["ac_sensor"]["left_x_offset"])
    self.right_sensor_x_offset  = float(conf["ac_sensor"]["right_x_offset"])
    self.calibration_y_offset_1 = float(conf["ac_sensor"]["calibration_y_offset_1"])
    self.calibration_y_offset_2 = float(conf["ac_sensor"]["calibration_y_offset_2"])
    self.calibration_x_offset   = float(conf["ac_sensor"]["calibration_x_offset"])
    self.sensor_distance        = self.field_width - self.left_sensor_x_offset + self.right_sensor_x_offset
    self.sonic_speed            = float(conf["ac_sensor"]["sonicspeed"])
    self.overhead_left          = float(conf["ac_sensor"]["overhead_left"])
    self.overhead_right         = float(conf["ac_sensor"]["overhead_right"])

    self.log_handler = logHandler.get_log_handler()
    self.log_handler.log_and_print("start acoustic sensor")

    # temporary calibration variables
    self.time_vals = [[],[]]
    self.cal_values = {
      "front": [0, 0],
      "back": [0, 0] 
    }
    self.n = 0

  def start(self):
    if not conn.isConnected():
      conn.open(port = self.conf["arduino"]["port"])
    conn.addRecvCallback(self._readCb)
    # generate dummy values until arduino is ready
    self.dummyActive = True
    dummyThread = threading.Thread(target=self._readCb_dummy)
    dummyThread.start()
    
  def start_calibration(self):
    self.calibration_state.reset_state()
    self.time_vals = [[],[]]
    self.calibration_state.next_state()

  def stop(self):
    self.log_handler.log_and_print("stop acoustic sensor")
    self.dummyActive = False
    conn.close()

  def _readCb_dummy(self):
    self.log_handler.log_and_print("acoustic sensor: generating test values")
    while self.dummyActive:
      if self.n % 4 < 1:
        dummyPosition = (0, self.n%1 * self.field_height)
      elif self.n % 4 < 2:
        dummyPosition = (self.n%1 * self.field_width, self.field_height)
      elif self.n % 4 < 3:
        dummyPosition = (self.field_width, self.field_height - self.n%1 * self.field_height)
      else:
        dummyPosition = (self.field_width - self.n%1 * self.field_width, 0)

      self.n += 0.01

      if self.calibration_state.get_state() in [self.calibration_state.ACCUMULATING_1, self.calibration_state.WAITING_POS_1]:
        dummyPosition = (self.calibration_x_offset + random.randint(-5,5), self.calibration_y_offset_1 + random.randint(-5,5))
        
      elif self.calibration_state.get_state() in [self.calibration_state.ACCUMULATING_2, self.calibration_state.WAITING_POS_2]:
        dummyPosition = (self.calibration_x_offset + random.randint(-5,5), self.calibration_y_offset_2 + random.randint(-5,5))

      # these dummy parameters should also appear after the calibration
      dummyValue = self.predict_values(dummyPosition, sonic_speed=0.35, overhead_left=30, overhead_right=100)

      self.calibrate(dummyValue)
      position = self.calculate_position(dummyValue)
      if position != None:
        self.pass_to_gui(position + dummyValue)
      time.sleep(0.012)
    self.log_handler.log_and_print("acoustic sensor: disabled test mode")

  def _readCb(self, raw):
    if self.dummyActive == True:
      self.dummyActive = False
    value = conn.getAcousticRTTs()
    # partially missing values will be ignored
    if value[0] >= 0 and value[1] >= 0:
      self.calibrate(value)
      position = self.calculate_position(value)
      if position != None:
        self.pass_to_gui(position + value)

  def calibrate(self, value):
    if self.calibration_state.get_state() == self.calibration_state.ACCUMULATING_1:
      self.time_vals[0].append(value[0])
      self.time_vals[1].append(value[1])
      self.calibration_state.progress = len(self.time_vals[0]) / 2
      if len(self.time_vals[0]) >= 100:
        self.cal_values["front"][0]  = statistics.mean(self.time_vals[0])
        self.cal_values["front"][1] = statistics.mean(self.time_vals[1])
        self.time_vals = [[],[]]
        self.calibration_state.next_state() # signal gui to get next position

    elif self.calibration_state.get_state() == self.calibration_state.ACCUMULATING_2:
      self.time_vals[0].append(value[0])
      self.time_vals[1].append(value[1])
      self.calibration_state.progress = 50 + len(self.time_vals[0]) / 2
      if len(self.time_vals[0]) >= 100:
        self.cal_values["back"][0]  = statistics.mean(self.time_vals[0])
        self.cal_values["back"][1] = statistics.mean(self.time_vals[1])

        # all values have been captured
        self.log_handler.log_and_print("calibration measurements:", self.cal_values)
        
        # calculate distances from config
        #                                     /|\                              
        #    d1    d2                     d3 / | \ d4                     
        #   _..'|'.._y_off + calYoff_2      /  |  \y_off + calYoff_2     
        #  /____|____\                     /___|___\                 
        # x_off                            x_off                     
        distance_1 = math.sqrt((self.calibration_x_offset + self.left_sensor_x_offset)**2 + (self.sensor_y_offset + self.calibration_y_offset_1)**2 )
        distance_2 = math.sqrt((self.calibration_x_offset + self.left_sensor_x_offset)**2 + (self.sensor_y_offset + self.calibration_y_offset_2)**2 )
        distancedif = distance_2 - distance_1
        timedif = self.cal_values["back"][0] - self.cal_values["front"][0]
        # speed of sound in mm/us
        sonicspeed_1 = distancedif / timedif

        # same for the second set of values
        distance_3 = math.sqrt((self.right_sensor_x_offset + (self.field_width - self.calibration_x_offset))**2 + (self.sensor_y_offset + self.calibration_y_offset_1)**2 )
        distance_4 = math.sqrt((self.right_sensor_x_offset + (self.field_width - self.calibration_x_offset))**2 + (self.sensor_y_offset + self.calibration_y_offset_2)**2 )
        distancedif = distance_4 - distance_3
        timedif = self.cal_values["back"][1] - self.cal_values["front"][1]
        sonicspeed_2 = distancedif / timedif

        # processing time overhead in us
        overhead_1 = statistics.mean((self.cal_values["front"][0] - distance_1/sonicspeed_1, self.cal_values["back"][0] - distance_2/sonicspeed_1))
        overhead_2 = statistics.mean((self.cal_values["front"][1] - distance_3/sonicspeed_2, self.cal_values["back"][1] - distance_4/sonicspeed_2))

        # calculate calibration results
        self.sonic_speed = statistics.mean((sonicspeed_1,sonicspeed_2))
        self.overhead_left  = overhead_1
        self.overhead_right = overhead_2

        self.log_handler.log_and_print("calibration results:")
        self.log_handler.log_and_print("  sonicspeed:     {:8.6f} mm/us".format(self.sonic_speed))
        self.log_handler.log_and_print("  overhead_left:  {:8.3f} us".format(self.overhead_left))
        self.log_handler.log_and_print("  overhead_right: {:8.3f} us".format(self.overhead_right))

        self.calibration_state.next_state()

  def read(self):
    value = conn.getAcousticRTTs()
    return value

  def calculate_position(self, values):
    val1, val2 = values
    val1 -= self.overhead_left
    val2 -= self.overhead_right
    distance_left = val1 * self.sonic_speed
    distance_right = val2 * self.sonic_speed
    # compute intersection of distance circles
    x = (self.sensor_distance**2 - distance_right**2 + distance_left**2) / (2*self.sensor_distance) + self.left_sensor_x_offset
    if distance_left**2 - x**2 >= 0:
      y = math.sqrt(distance_left**2 - x**2) - self.sensor_y_offset
      return (x, y)
    else:
      return None

  def predict_values(self, position, sonic_speed, overhead_left, overhead_right):
    x, y = position
    distance_left = math.sqrt((y + self.sensor_y_offset)**2 + x**2)
    distance_right = math.sqrt(self.sensor_distance**2 + distance_left**2 - x * 2 * self.sensor_distance - x * self.left_sensor_x_offset)
    return (distance_left / sonic_speed + overhead_left, distance_right / sonic_speed + overhead_right)

  def pass_to_gui(self, data):
    self.queue.put(("data", data))