SafeCast: Risk-Responsive Motion Forecasting for Autonomous Vehicles

2025-03-28

AI Summary

Existing autonomous driving motion prediction methods lack explicit safety constraints, hindering accurate modeling of complex interactions among traffic agents, environmental context, and dynamic risks—thereby compromising system safety and reliability. To address this, we propose a risk-responsive motion prediction model that, for the first time, embeds Responsibility-Sensitive Safety (RSS) principles directly into the prediction framework, enabling joint optimization of safety awareness and motion forecasting. We further introduce a Graph Uncertainty Feature (GUF) module within a graph attention network, incorporating learnable stochastic noise to enhance cross-scenario generalization and robustness. The model is trained and evaluated on four large-scale real-world traffic datasets—NGSIM, HighD, ApolloScape, and MoCAD—achieving state-of-the-art (SOTA) accuracy across all benchmarks. Moreover, its lightweight architecture and low inference latency enable real-time onboard deployment.

Technology Category

• Computer Vision: Motion & Tracking
• Natural Language Processing: Safety and Robustness
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Accurate motion forecasting is essential for the safety and reliability of autonomous driving (AD) systems. While existing methods have made significant progress, they often overlook explicit safety constraints and struggle to capture the complex interactions among traffic agents, environmental factors, and motion dynamics. To address these challenges, we present SafeCast, a risk-responsive motion forecasting model that integrates safety-aware decision-making with uncertainty-aware adaptability. SafeCast is the first to incorporate the Responsibility-Sensitive Safety (RSS) framework into motion forecasting, encoding interpretable safety rules--such as safe distances and collision avoidance--based on traffic norms and physical principles. To further enhance robustness, we introduce the Graph Uncertainty Feature (GUF), a graph-based module that injects learnable noise into Graph Attention Networks, capturing real-world uncertainties and enhancing generalization across diverse scenarios. We evaluate SafeCast on four real-world benchmark datasets--Next Generation Simulation (NGSIM), Highway Drone (HighD), ApolloScape, and the Macao Connected Autonomous Driving (MoCAD)--covering highway, urban, and mixed-autonomy traffic environments. Our model achieves state-of-the-art (SOTA) accuracy while maintaining a lightweight architecture and low inference latency, underscoring its potential for real-time deployment in safety-critical AD systems.

Problem

• Integrate safety constraints into autonomous vehicle motion forecasting
• Model complex interactions among traffic agents and environment
• Enhance robustness with uncertainty-aware adaptability in diverse scenarios

Innovation

• Integrates RSS framework for safety rules
• Uses Graph Uncertainty Feature for robustness
• Achieves SOTA accuracy with lightweight design