Vehicle Lane-Changes Trajectory Prediction Model Considering External Parameters

The ability to predict the motion of vehicles is essential for autonomous vehicles. Aiming at problem that existing models cannot make full use external parameters including outline and lane, we proposed a model thoroughly when predicting trajectory in straight-line non-free flow state. Meanwhile, dynamic sensitive area filter out inconsequential surrounding historical their are used as inputs. A shared Long Short-Term Memory (LSTM) cell encode explicit states obtained by mapping parameters. hidden from last step extract latent driving intent. Then, convolution layer designed fuse feed into next prediction circle decoder decode trajectory. experiment result shows can shorten training time 75.86% state-of-the-art work. Compared with other models, accuracy our improved 23.7%. model's anti-interference also improved.