A new method for semantic consistency verification of aviation radiotelephony communication based on LSTM-RNN

Abstract

In Aviation Radiotelephony Communication (ARC), the incorrect readback between pilots and air traffic controllers has a vital effect on aircraft flight safety. To make aircraft safer in aviation, International Civil Aviation Organization (ICAO) has improved the communication standard of air traffic. However, the accidents caused by incorrect readback of ARC still happen unavoidably. To reduce the risk of incorrect readback, this paper proposes a method verifying the semantic consistency of the ARC. We firstly apply Recurrent Neural Network (RNN) and Long Short-Term memory Recurrent Neural Network (LSTM-RNN) to extract the semantic meaning of ARC and represent it with semantic vector, and then add a sigmoid layer at the output of RNN or LSTM-RNN to verify the semantic consistency. The RNN or LSTM-RNN are trained in a supervised learning method. We evaluate the proposed architecture on ARC corpus. The experimental results show that the proposed method is effective in semantic consistency verification of ARC, and LSTM-RNN outperforms the RNN in this task.