Beyond Features: How Dataset Design Influences Multi-Agent Trajectory Prediction Performance

2025-07-07

AI Summary

This study systematically investigates how dataset design affects multi-agent trajectory prediction performance, focusing on feature efficacy, cross-dataset transferability, and geographic diversity. We introduce a custom L4-level motion prediction dataset featuring enhanced map representations and agent-centric attributes, and conduct cross-dataset experiments and country-wise evaluations on the Argoverse 2 benchmark. Our findings are threefold: (1) State-of-the-art models achieve high accuracy using only minimal base features; adding hand-crafted features yields no significant improvement—demonstrating that existing public datasets already provide sufficient representational capacity. (2) Models exhibit robust knowledge transfer across geographically distinct regions (e.g., the U.S., Germany, Japan), indicating that driving behavior shares more universal patterns than culture-specific ones. (3) We provide the first empirical evidence that geographic diversity in training data critically enhances model generalization. These results establish theoretical foundations and practical guidelines for principled trajectory prediction dataset construction.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Multiagent Systems: Adversarial Agents

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Accurate trajectory prediction is critical for safe autonomous navigation, yet the impact of dataset design on model performance remains understudied. This work systematically examines how feature selection, cross-dataset transfer, and geographic diversity influence trajectory prediction accuracy in multi-agent settings. We evaluate a state-of-the-art model using our novel L4 Motion Forecasting dataset based on our own data recordings in Germany and the US. This includes enhanced map and agent features. We compare our dataset to the US-centric Argoverse 2 benchmark. First, we find that incorporating supplementary map and agent features unique to our dataset, yields no measurable improvement over baseline features, demonstrating that modern architectures do not need extensive feature sets for optimal performance. The limited features of public datasets are sufficient to capture convoluted interactions without added complexity. Second, we perform cross-dataset experiments to evaluate how effective domain knowledge can be transferred between datasets. Third, we group our dataset by country and check the knowledge transfer between different driving cultures.

Problem

• Impact of dataset design on trajectory prediction accuracy
• Effect of feature selection and geographic diversity
• Cross-dataset knowledge transfer in multi-agent settings

Innovation

• Enhanced map and agent features in dataset
• Cross-dataset transfer of domain knowledge
• Geographic diversity for driving culture analysis