Motion Forecasting for Autonomous Vehicles: A Survey

2025-02-10

AI Summary

This survey systematically reviews advances in traffic participant motion prediction for autonomous vehicles. We formally define the problem under two paradigms—scene-driven and perception-driven—and unify the taxonomy into supervised and self-supervised learning approaches. Methodologically, we structurally analyze key components, categorizing mainstream techniques including LSTMs, graph neural networks (GNNs), Transformers, contrastive learning, and generative modeling, while summarizing core challenges, benchmark datasets, and evaluation metrics. Our contribution is the most comprehensive motion prediction technology map to date, explicitly delineating performance boundaries and application scopes of diverse methods. This work provides an authoritative reference for algorithm selection, novel model design, and enhancing decision-making safety and generalization in autonomous vehicles. We further propose a structured framework outlining future research directions, including causal reasoning, uncertainty-aware prediction, and cross-modal generalization.

Technology Category

• Computer Vision: Motion & Tracking
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Unsupervised & Self-Supervised Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

In recent years, the field of autonomous driving has attracted increasingly significant public interest. Accurately forecasting the future behavior of various traffic participants is essential for the decision-making of Autonomous Vehicles (AVs). In this paper, we focus on both scenario-based and perception-based motion forecasting for AVs. We propose a formal problem formulation for motion forecasting and summarize the main challenges confronting this area of research. We also detail representative datasets and evaluation metrics pertinent to this field. Furthermore, this study classifies recent research into two main categories: supervised learning and self-supervised learning, reflecting the evolving paradigms in both scenario-based and perception-based motion forecasting. In the context of supervised learning, we thoroughly examine and analyze each key element of the methodology. For self-supervised learning, we summarize commonly adopted techniques. The paper concludes and discusses potential research directions, aiming to propel progress in this vital area of AV technology.

Problem

• Forecasting traffic participants' behavior
• Formulating motion forecasting problem
• Classifying supervised and self-supervised learning

Innovation

• scenario-based motion forecasting
• perception-based motion forecasting
• self-supervised learning techniques