Vehicle track prediction method based on intention perception space-time attention network

The invention discloses a vehicle track prediction method based on an intention perception space-time attention network, and relates to the technical field of intelligent vehicles. The method comprises the following steps: step 1, offline training of a vehicle trajectory prediction model: constructing a training database and training the vehicle trajectory prediction model through acquisition and processing of a large-scale real driving scene; and 2, on-line real-time prediction of the vehicle track prediction model: after real-time data acquisition and data preprocessing of the vehicle, the trained vehicle track prediction model is used for on-line real-time prediction, and the track of the target vehicle can be predicted. In consideration of the coupling between the driving intention and the neighbor vehicle interaction, an intention attention mechanism is provided, and attention weights are distributed in the time dimension to extract historical information of the vehicle; an interaction relation capturing