Understanding and Detecting Platform-Specific Violations in Android Auto Apps
Despite over 3.5 million Android apps and 200+ million Android Auto-compatible vehicles, only a few hundred apps support Android Auto due to platform-specific compliance requirements. Android Auto mandates service-based architectures in which the vehicle system invokes app callbacks to render the UI and handle interactions, which is fundamentally different from standard Activity-based Android development. Through an empirical study analysis of 98 issues across 14 Android Auto app repositories, we identified three major compliance failure categories: media playback errors, UI rendering issues, and voice command integration failures in line with mandatory requirements for integrating Android Auto support. We introduce AutoComply, a static analysis framework capable of detecting these compliance violations through the specialized analysis of platform-specific requirements. AutoComply constructs a Car-Control Flow Graph (CCFG) extending traditional control flow analysis to model the service-based architecture of Android Auto apps. Evaluating AutoComply on 31 large-scale open-source apps, it detected 27 violations (13x more than Android Lint) with zero false positives, achieving 2x faster analysis. Developers have acknowledged 14 of these violations with 8 fixes already implemented, validating AutoComply’s practical effectiveness.
Tue 14 AprDisplayed time zone: Brasilia, Distrito Federal, Brazil change
11:00 - 12:30 | Session 6: Testing Around the WorldAST 2026 at Oceania VI Chair(s): Hokeun Kim Arizona State University | ||
11:00 30mTalk | Understanding and Detecting Platform-Specific Violations in Android Auto Apps AST 2026 Pre-print Media Attached | ||
11:30 30mTalk | A Unified Benchmark for Out-of-Distribution Detection for Autonomous Driving Systems AST 2026 Xiangyu Li SeysoAI, Jingyu ZHANG Hong Kong Metropolitan University, Jacky Keung City University of Hong Kong, Xiaoxue Ma Hong Kong Metropolitan University, Yihan Liao City University of Hong Kong Pre-print Media Attached | ||
12:00 30mTalk | HYDRA: A Hybrid Heuristic-Guided Deep Representation Architecture for Predicting Latent Zero-Day Vulnerabilities in Patched Functions AST 2026 Mohammad Farhad University of Louisiana at Lafayette, Sabbir Rahman University of Louisiana at Lafayette, Shuvalaxmi Dass University of Louisiana at Lafayette Pre-print Media Attached | ||