Adaptive Navigation Strategy for Low‐Cost IMU‐Assisted Vehicles in GNSS‐Denied Traffic Environment

In the context of increasingly complex and diverse traffic environments, where intelligent and connected vehicles (ICVs) coexist with conventional human‐driven vehicles, maintaining reliable navigation under global navigation satellite system (GNSS) outages is crucial for supporting adaptive driving strategies and ensuring operational safety. Low‐cost inertial measurement units (IMUs) offer promising solutions due to their low computational load and high self‐sufficiency, yet error accumulation remains a persistent challenge, particularly in real‐world mixed traffic scenarios. This study introduces a dilated convolutional neural network (DCN)–driven framework to directly estimate vehicle forward velocity and IMU error parameters from raw IMU measurements, addressing the reliance on hardware‐based odometry by extending nonholonomic constraints (NHC) into three‐dimensional velocity constraints. By dynamically optimizing IMU error parameters through integration with an error model, the proposed method mitigates the adverse effects of inherent noise in low‐cost IMUs, enabling robust navigation in GNSS‐denied environments. Validation using a GNSS/INS dataset demonstrates that the approach accurately estimates vehicle position while significantly suppressing error accumulation, which is pivotal for maintaining reliable navigation in heterogeneous traffic flows where autonomous and human‐driven vehicles coexist. This contributes to the development of robust vehicle autonomy and enhanced safety in mixed‐traffic ecosystems, enabling more adaptive and resilient driving strategies.