Evaluation of Differentially Constrained Motion Models for Graph-Based Trajectory Prediction
This work addresses the need for physically feasible and interpretable trajectory predictions in autonomous driving, though it is incremental as it builds on an existing graph-neural-network model.
The study evaluated differentially constrained motion models combined with graph-based neural networks for autonomous driving trajectory prediction, finding that simpler low-order integrator models outperform more complex kinematic models and that second-order numerical solvers like Heun's method significantly improve accuracy over first-order methods like Euler forward.
Given their flexibility and encouraging performance, deep-learning models are becoming standard for motion prediction in autonomous driving. However, with great flexibility comes a lack of interpretability and possible violations of physical constraints. Accompanying these data-driven methods with differentially-constrained motion models to provide physically feasible trajectories is a promising future direction. The foundation for this work is a previously introduced graph-neural-network-based model, MTP-GO. The neural network learns to compute the inputs to an underlying motion model to provide physically feasible trajectories. This research investigates the performance of various motion models in combination with numerical solvers for the prediction task. The study shows that simpler models, such as low-order integrator models, are preferred over more complex, e.g., kinematic models, to achieve accurate predictions. Further, the numerical solver can have a substantial impact on performance, advising against commonly used first-order methods like Euler forward. Instead, a second-order method like Heun's can greatly improve predictions.