def get_all_differences(all_sensor_data, all_ground_truths):

    pxs, pys, vxs, vys, rhos, phis, drhos = [], [], [], [], [], [], []

    for s, t in zip(all_sensor_data, all_ground_truths):

        if s.get_name() == 'lidar':

            spx, spy, _, _ = s.get()
            tpx, tpy, _, _ = t.get()

            pxs += [spx - tpx]
            pys += [spy - tpy]

        else:

            spx, spy, svx, svy = s.get()
            tpx, tpy, tvx, tvy = t.get()

            srho, sphi, sdrho = s.get_raw()
            trho, tphi, tdrho = cartesian_to_polar(tpx, tpy, tvx, tvy)

            pxs += [spx - tpx]
            pys += [spy - tpy]
            vxs += [svx - tvx]
            vys += [svy - tvy]
            rhos += [srho - trho]
            phis += [sphi - tphi]
            drhos += [sdrho - tdrho]

    return pxs, pys, vxs, vys, rhos, phis, drhos
Example #2
0
def kalman(data, kalmanFilter, timestamp):
    dt = time_difference(timestamp, data.get_timestamp())
    timestamp = data.get_timestamp()

    kalmanFilter.F[0, 2], kalmanFilter.F[1, 3] = dt, dt
    kalmanFilter.Q = updateQ(dt)
    kalmanFilter.predict()

    z = np.matrix(data.get_raw()).T
    x = kalmanFilter.x

    if data.get_name() == 'radar':

        px, py, vx, vy = x[0, 0], x[1, 0], x[2, 0], x[3, 0]
        rho, phi, drho = cartesian_to_polar(px, py, vx, vy)
        H = calculate_jacobian(px, py, vx, vy)
        Hx = (np.matrix([[rho, phi, drho]])).T
        R = radar_R

    elif data.get_name() == 'lidar':

        H = lidar_H
        Hx = lidar_H * x
        R = lidar_R

    kalmanFilter.update(z, H, Hx, R)

    # if data.get_name() == 'lidar':
    # 	print x, kalmanFilter.x, "lidar"
    # elif data.get_name() == 'radar':
    # 	print x, kalmanFilter.x, "radar"

    return timestamp
    def update(self, data):

        dt = time_difference(self.timestamp, data.get_timestamp())
        self.timestamp = data.get_timestamp()

        self.kalmanFilter.updateF(dt)
        self.updateQ(dt)
        self.kalmanFilter.predict()

        z = np.matrix(data.get_raw()).T
        x = self.kalmanFilter.getx()

        if data.get_name() == 'radar':

            px, py, vx, vy = x[0, 0], x[1, 0], x[2, 0], x[3, 0]
            rho, phi, drho = cartesian_to_polar(px, py, vx, vy)
            H = calculate_jacobian(px, py, vx, vy)
            Hx = (np.matrix([[rho, phi, drho]])).T
            R = self.radar_R

        elif data.get_name() == 'lidar':

            H = self.lidar_H
            Hx = self.lidar_H * x
            R = self.lidar_R

        self.kalmanFilter.update(z, H, Hx, R)
Example #4
0
def parse_data(file_path, ENCODE=False):
    """
    Args:
    file_path 
      - path to a text file with all data. 
      - each line should have the following format:
          [SENSOR ID] [SENSOR RAW VALUES] [TIMESTAMP] [GROUND TRUTH VALUES]
          Whereas radar has three measurements (rho, phi, rhodot), lidar has two measurements (x, y).

          Specifically: 
          
          For a row containing radar data, the columns are: 
          sensor_type, rho_measured, phi_measured, rhodot_measured, timestamp, x_groundtruth, y_groundtruth, vx_groundtruth, vy_groundtruth
          
          For a row containing lidar data, the columns are:
          sensor_type, x_measured, y_measured, timestamp, x_groundtruth, y_groundtruth, vx_groundtruth, vy_groundtruth
          
          Example 1: 
          line with three measurements from a radar sensor in polar coordinate 
          followed by a timestamp in unix time 
          followed by the the "ground truth" which is
          actual real position and velocity in cartesian coordinates (four state variables)

          R 8.46642 0.0287602 -3.04035  1477010443399637  8.6 0.25  -3.00029  0
          (R) (rho) (phi) (drho) (timestamp) (real x) (real y) (real vx) (real vy)

          Example 2:
          line with two measurements from a lidar sensor in cartesian coordinates 
          followed by a timestamp in unix time
          followed by the the "ground truth" which is 
          the actual real position and velocity in cartesian coordinates (four state variables)

          L 8.44818 0.251553  1477010443449633  8.45  0.25  -3.00027  0

    Returns:
      all_sensor_data, all_ground_truths
      - two lists of DataPoint() instances 

  """

    all_sensor_data = []
    all_ground_truths = []

    if (ENCODE):
        message = 0

    with open(file_path) as f:

        for line in f:
            data = line.split()

            if data[0] == 'L':

                data_read = {
                    'timestamp': int(data[3]),
                    'name': 'lidar',
                    'x': float(data[1]),
                    'y': float(data[2]),
                    'vx': 0.0,
                    'vy': 0.0
                }
                # print('helpers:parse_data: data read LiDAR', data_read['x'],data_read['y'],data_read['vx'],data_read['vx'])
                sensor_data = DataPoint(data_read)
                # print('helpers:parse_data: data read LiDAR', sensor_data.data[0],sensor_data.data[1],sensor_data.data[2],sensor_data.data[3] )
                g = {
                    'timestamp': int(data[3]),
                    'name': 'state',
                    'x': float(data[4]),
                    'y': float(data[5]),
                    'vx': float(data[6]),
                    'vy': float(data[7])
                }

                ground_truth = DataPoint(g)
                # print('helpers:parse_data: Ground truth LiDAR', ground_truth.data[0],ground_truth.data[1], ground_truth.data[2], ground_truth.data[3] )

            elif data[0] == 'R':
                #herer sensor_data is the output of the DataPoint class
                if (ENCODE):  #Here we implement the QIM if used as an option
                    #convert to cartesian
                    x, y, vx, vy = polar_to_cartesian(float(data[1]),
                                                      float(data[2]),
                                                      float(data[3]))
                    #embedd data - only modify x,y
                    x, y = QIM_encode_twobit(np.array([x, y]), message)
                    #print('helpers:parse_data: encoded data read raDAR', sensor_data.data[0],sensor_data.data[1],sensor_data.data[2],sensor_data.data[3] )
                    message += 1
                    message %= 4
                    #convert to polar
                    rhho, phhi, derho = cartesian_to_polar(x, y, vx, vy)
                    data_read = {
                        'timestamp': int(data[4]),
                        'name': 'radar',
                        'rho': float(rhho),
                        'phi': float(phhi),
                        'drho': float(derho)
                    }
                else:
                    data_read = {
                        'timestamp': int(data[4]),
                        'name': 'radar',
                        'rho': float(data[1]),
                        'phi': float(data[2]),
                        'drho': float(data[3])
                    }
                sensor_data = DataPoint(data_read)
                #print('helpers:parse_data: data read raDAR', sensor_data.data[0],sensor_data.data[1],sensor_data.data[2],sensor_data.data[3] )

                g = {
                    'timestamp': int(data[4]),
                    'name': 'state',
                    'x': float(data[5]),
                    'y': float(data[6]),
                    'vx': float(data[7]),
                    'vy': float(data[8])
                }

                ground_truth = DataPoint(g)
                # print('helpers:parse_data: Ground truth raDAR',ground_truth.data[0],ground_truth.data[1], ground_truth.data[2], ground_truth.data[3]  )

            all_sensor_data.append(sensor_data)
            all_ground_truths.append(ground_truth)

    return all_sensor_data, all_ground_truths