GAMDTP: Dynamic Trajectory Prediction with Graph Attention Mamba Network

TL;DR

GAMDTP addresses dynamic trajectory prediction for autonomous driving by integrating Graph Attention Networks with Mamba-SSM through a gate-based fusion, enabling balanced local and global feature extraction. The method leverages an HPNet-inspired backbone to encode HD maps and historical trajectories, and introduces a Quality Scoring Mechanism to coordinate proposal and refinement stages in a two-stage forecast. Empirical results on Argoverse and INTERACTION demonstrate state-of-the-art performance across marginal and joint metrics, with improved efficiency and robust multimodal predictions. This approach advances real-time trajectory forecasting by combining accuracy with computational efficiency, benefiting planning and safety in autonomous systems.

Abstract

Accurate motion prediction of traffic agents is crucial for the safety and stability of autonomous driving systems. In this paper, we introduce GAMDTP, a novel graph attention-based network tailored for dynamic trajectory prediction. Specifically, we fuse the result of self attention and mamba-ssm through a gate mechanism, leveraging the strengths of both to extract features more efficiently and accurately, in each graph convolution layer. GAMDTP encodes the high-definition map(HD map) data and the agents' historical trajectory coordinates and decodes the network's output to generate the final prediction results. Additionally, recent approaches predominantly focus on dynamically fusing historical forecast results and rely on two-stage frameworks including proposal and refinement. To further enhance the performance of the two-stage frameworks we also design a scoring mechanism to evaluate the prediction quality during the proposal and refinement processes. Experiments on the Argoverse dataset demonstrates that GAMDTP achieves state-of-the-art performance, achieving superior accuracy in dynamic trajectory prediction.

Figures (3)

• Figure 1: Overview of GAMDTP. The encoder processes raw input features such as HD map and agent trajectory information. Our proposed Graph Attention Mamba module is applied in the components Agent GAM, Historical Prediction GAM and Mode GAM, which extracts spatio-temporal features. Decoder generates the final predicted trajectories and probability and the score decoder further evaluates and prioritizes trajectory candidates for refinement through generate a score for each result, ensuring accurate and reliable predictions.
• Figure 2: Our proposed Graph Attention Mamba module, which integrates Mamaba block and graph attention block. The input features include node features and edge features, which first normalized through a Layernorm(LN) layer before processed by Mamba and GAT blocks. The output from these blocks are fused using a gate mechanism, where the sigmoid function dynamically generates a gate signal G to balance their contributions.
• Figure 3: Comparison our GAMDTP with baseline.