A Multidimensional Graph Fourier Transformation Neural Network for Vehicle Trajectory Prediction
This work addresses trajectory prediction for autonomous vehicles on highways, representing an incremental improvement over existing graph neural network methods.
The paper tackles long-term vehicle trajectory prediction on highways by introducing the multidimensional Graph Fourier Transformation Neural Network (GFTNN), which uses a spectral representation to aggregate scenario properties without recurrent elements, and it outperforms state-of-the-art models on highD and NGSIM datasets.
This work introduces the multidimensional Graph Fourier Transformation Neural Network (GFTNN) for long-term trajectory predictions on highways. Similar to Graph Neural Networks (GNNs), the GFTNN is a novel network architecture that operates on graph structures. While several GNNs lack discriminative power due to suboptimal aggregation schemes, the proposed model aggregates scenario properties through a powerful operation: the multidimensional Graph Fourier Transformation (GFT). The spatio-temporal vehicle interaction graph of a scenario is converted into a spectral scenario representation using the GFT. This beneficial representation is input to the prediction framework composed of a neural network and a descriptive decoder. Even though the proposed GFTNN does not include any recurrent element, it outperforms state-of-the-art models in the task of highway trajectory prediction. For experiments and evaluation, the publicly available datasets highD and NGSIM are used