High-Fidelity simulator for offline Pre-Validation of agricultural tractor navigation plans based on An optimized Multi-Stage steering model

Pre-validating agricultural tractor navigation plans is costly and risky when it relies on repeated on-site debugging with multi-ton machinery. This study develops an offline simulator, termed the High‑Fidelity Simulator (HFS), to support offline pre-validation and screening of navigation plans by reproducing accurate trajectory-level motion under a fixed velocity assumption and a specific tractor and controller configuration. Here, fidelity is quantified using trajectory reproduction error metrics, including Average, RMSE and Maximum trajectory differences and navigation tracking errors. The core technical contribution in this study is a multi-stage steering simulator with parameters optimization, which improves steering response matching and trajectory reproduction accuracy over a traditional first-order delay steering model. The HFS further includes farm terrain reconstruction, a basic tractor model, Global Navigation Satellite System (GNSS) error simulation, and a surface-specific empirical turning radius correction. In step response tests on asphalt and soil conditions, the HFS achieved satisfactory trajectory reproduction accuracy and showed advantages in challenging soil cases compared with the traditional first-order delay steering baseline. In real navigation test, the HFS produced navigation-level similarity with comparable navigation tracking errors to data logged from physical space. These results demonstrate accurate offline trajectory reproduction for the tested tractor and controller setting, suggesting that the proposed HFS can serve as a practical tool for offline screening of navigation plans under the tested use case to reduce costly and risky on-site debugging prior to field deployment.