Code and data for "Clearance-Verified Spline Smoothing for Sampling-Based Navigation: Planning Margin, Reference Curvature and LQR/MPC Tracking"
Year: 2026
Type: software
Abstract: Code, raw results and figures for an empirical study of how the clearance budget (ε) of a sampling-based planner and the spline-smoothing setting affect LQR and MPC path tracking on a kinematic bicycle model. The package contains the experiment code, the random-map campaigns (20 seeds, 6 values of ε, two smoothing settings), the structured-map campaign (narrow passage and dead end), the scripts that produce the tables and figures, and a script (reproduce.sh) that reruns the experiments from the stated seeds and compares the output with the shipped results. Results depend on the library versions in requirements.txt (notably SciPy 1.18.0). A rerun on a second machine (Windows) reproduced the random-map campaigns exactly; for the structured-map campaign, planning and clearance were identical and the tracking metrics agreed only approximately (see reproduction_windows/COMPARISON.txt). Licences: code MIT, data and figures CC BY 4.0.
Author: BUCAK, İhsan Ömür
Institution: Iğdır Üniversitesi
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7221317295