Hierarchical Fault Localization for Autonomous Driving Systems with Hypothesis Validation and Intent Analysis

2026-07-14

AI Summary

Current testing methodologies for autonomous driving systems struggle to precisely trace behavioral failures back to their root causes in the source code, resulting in debugging processes that heavily rely on manual effort. To address this challenge, this work proposes HINT, a novel framework that integrates hypothesis validation with design-intent analysis to enable end-to-end hierarchical fault localization. In the first stage, HINT identifies faulty modules through multimodal execution abstraction and causal reasoning; in the second stage, it reconstructs the intended design logic and performs reliability-aware consistency checks to pinpoint suspicious code without requiring re-simulation. Experiments on the Apollo autonomous driving system demonstrate that HINT significantly outperforms baseline approaches in both module-level diagnosis and code-level localization, achieving a 77.8% end-to-end Class@5 accuracy on real-world defects and substantially reducing manual debugging overhead.

Abstract

Comprehensive testing is essential for the safety and reliability of Autonomous Driving Systems (ADS). Existing techniques can detect system-level failures or attribute them to coarse-grained modules, but they often fall short of localizing the root cause in source code. As a result, debugging remains labor-intensive, requiring developers to connect behavioral violations with complex implementation logic. To address this gap, we present HINT, a two-phase framework for hierarchical ADS fault localization based on hypothesis validation and intent analysis. In Phase I, HINT transforms failure-triggering execution recordings into multi-modal abstractions and uses causal reasoning to identify the responsible module. In Phase II, it reconstructs design-side intent and implementation-side behavior, then localizes suspicious code through reliability-aware consistency checking, without costly re-simulation. We evaluate HINT on Apollo across diverse failure modes and modules. The results show that HINT achieves the strongest overall performance across module-level diagnosis and code-level localization metrics, with 77.8% end-to-end Class@5 accuracy on real-world bugs.

Problem

• fault localization
• autonomous driving systems
• root cause analysis
• debugging
• code-level diagnosis

Innovation

• hierarchical fault localization
• hypothesis validation
• intent analysis
• causal reasoning
• reliability-aware consistency checking