Pykalman AdditiveUnscentedKalmanFilter Initialization

Question

I am attempting to use the AdditiveUnscentedKalman filter from pykalman. I am given noisy x,y values, and I am trying to get the x,y,xdot,ydot out of the filter.

I am getting values out of the filter, but it is not what I expect. State: [[ 481.65737052 477.23904382 0. 0. ] [ 659.29999618 659.28265402 58.33365188 59.77883149]] Obs: [478, 660, -0.4666666666666667, -0.36666666666666664, -2.4756234162106834, 1.2145259141964182]

The obs[0] and obs[1] are the measured x y and state is what is coming out of the filter. state[0][0] and state[1][0] look to be an x,y and state[0][1] and state[1][1] seem to be an x,y. I have no idea what they other numbers are supposed to be by they are not acceptable velocities.

If someone could just validate that I have am using the correct transition function it would be greatly appreciated.

transition_covariance = np.array([[100.0, 0.0, 0.0, 0.0],
                                         [0.0, 100.0, 0.0, 0.0],
                                         [0.0, 0.0, 100.0, 0.0],
                                         [0.0, 0.0, 0.0, 100.0]])

observation_covariance = np.array([[0.4, 0.0],
                                       [0.0, 0.4]])

initial_state_covariance = np.array([[100.0, 0.0, 0.0, 0.0],
                                         [0.0, 100.0, 0.0, 0.0],
                                         [0.0, 0.0, 1000.0, 0.0],
                                         [0.0, 0.0, 0.0, 1000.0]])

self.ukf = AdditiveUnscentedKalmanFilter(transition_functions = self.transition_function,
                                             observation_functions = self.observation_function,
                                             transition_covariance = transition_covariance,
                                             observation_covariance = observation_covariance,
                                             initial_state_mean = initial_conditions,
                                             initial_state_covariance = initial_state_covariance)

def get_states(self, observations):
    return self.ukf.filter(observations)

# [x, y, xvel, yvel]
def transition_function(self, state):
    return np.array([state[0] + state[2] * self.dt,
                     state[1] + state[3] * self.dt,
                     state[2],
                     state[3]])

def observation_function(self, state):
    om = np.array([[1.0, 0.0, 0.0, 0.0],
                   [0.0, 1.0, 0.0, 0.0]])
    return np.matmul(om, state)

I am confused because the output of the .filter call is a matrix of 2X4 where I would expect a 1X4.