Hybrid Feature Coupled BiLSTM to Predict the Trajectories and Motion of the Autonomous Vehicles

Safety is a major research concern in the Autonomous Vehicle (AV) to avoid accidents on the road while driving the AVs. Due to minimal errors in the existing research, several accidents are unavoidable across the world. The Challenges such as Safety-critical decision-making in edge cases, and unexpected scenarios are employed in the existing research of AVs. To overcome the challenges in the research of trajectory and motion prediction, the Hybrid Features coupled BiLSTM (HF coupled BiLSTM) is proposed in the research. The research models aid in achieving the accurate prediction of the trajectories with the in-built feature maps such as the convolutional maps and the Spatiotemporal feature maps. The performance of the Hybrid feature coupled BiLSTM model is better as compared to the existing models of RNN, Attention LSTM, Spatio-temporal LSTM, LSTM-RNN, and Distributed discriminator-based BiLSTM for the trajectory and motion prediction in this research concerning the performance metrics with Mean Square Error (MSE) is 5.73, Root Mean Square Error (RMSE) is 2.39, and the Mean Absolute Error (MAE) is 1.83 for the NGSIM database.