TPK: Trustworthy Trajectory Prediction Integrating Prior Knowledge For Interpretability and Kinematic Feasibility

2025-05-10

AI Summary

Existing trajectory prediction models for autonomous driving suffer from three key limitations: physically infeasible outputs, opaque interaction reasoning, and poor generalization across heterogeneous traffic agents (e.g., vehicles, pedestrians, cyclists). To address these, we propose a unified multi-agent modeling framework featuring a novel differentiable physical projection mechanism that jointly incorporates social-force priors and class-specific dynamical constraints. We further design the DG-SFM module to yield interpretable, interaction-aware importance scores, and introduce differentiated kinematic models—including a new pedestrian dynamics submodel. Integrated within the HPTR architecture, our approach is rigorously evaluated on Argoverse 2. Results demonstrate complete elimination of physically implausible trajectories, significantly enhanced interpretability—evidenced by strong correlation between erroneous predictions and prior deviations—and competitive accuracy under substantially improved prediction reliability.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Planning, Routing, and Scheduling: Model-Based 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: Agentic search

Abstract

Trajectory prediction is crucial for autonomous driving, enabling vehicles to navigate safely by anticipating the movements of surrounding road users. However, current deep learning models often lack trustworthiness as their predictions can be physically infeasible and illogical to humans. To make predictions more trustworthy, recent research has incorporated prior knowledge, like the social force model for modeling interactions and kinematic models for physical realism. However, these approaches focus on priors that suit either vehicles or pedestrians and do not generalize to traffic with mixed agent classes. We propose incorporating interaction and kinematic priors of all agent classes--vehicles, pedestrians, and cyclists with class-specific interaction layers to capture agent behavioral differences. To improve the interpretability of the agent interactions, we introduce DG-SFM, a rule-based interaction importance score that guides the interaction layer. To ensure physically feasible predictions, we proposed suitable kinematic models for all agent classes with a novel pedestrian kinematic model. We benchmark our approach on the Argoverse 2 dataset, using the state-of-the-art transformer HPTR as our baseline. Experiments demonstrate that our method improves interaction interpretability, revealing a correlation between incorrect predictions and divergence from our interaction prior. Even though incorporating the kinematic models causes a slight decrease in accuracy, they eliminate infeasible trajectories found in the dataset and the baseline model. Thus, our approach fosters trust in trajectory prediction as its interaction reasoning is interpretable, and its predictions adhere to physics.

Problem

• Enhances trajectory prediction trustworthiness for mixed traffic agents
• Improves interpretability of agent interactions with rule-based scoring
• Ensures kinematic feasibility across diverse agent classes

Innovation

• Class-specific interaction layers for mixed agents
• DG-SFM rule-based interaction importance score
• Novel pedestrian kinematic model for feasibility