Method for Evaluating Vehicle Maneuverability Based on Real-World Routes

Sammanfattning

Scania’s existing circular maneuver evaluation tool is primarily designed to support standardized metrics. To further broaden its applicability to real-world driving scenarios, there is a need to extend the system’s capability to handle arbitrary routes based on actual geographical data. This thesis presents a method for evaluating vehicle maneuverability using real-world route information. The study first investigates B-spline fitting for route representation but identifies significant limitations in preserving local geometric details. To address this, a geometry-based approach is proposed that directly incorporates the vehicle’s turning radii into the analysis. The method comprises three main stages: Sharp Turn Detection, Curve Edge Generation, and Constraint Feasibility Check. In the final step, three critical constraints are formulated to determine whether a vehicle can successfully navigate a given curve. The algorithm was implemented in multiple programming languages, both for research convenience and for integration into operational systems. Validation was conducted on Scania’s test track, showing consistency between predicted and actual maneuverability outcomes. Performance evaluations on an Amazon Elastic Compute Cloud instance further demonstrated the method’s efficiency, real-time applicability, and robustness under extreme conditions. The study concludes that the proposed method is effective, computationally efficient, and practical for real-world deployment onto Scania’s maneuverability evaluation tool.