Certified Lower-Bound Orientation Search under Cross-Stage Objective Mismatch in Agricultural Coverage Path Planning
Year: 2026
Type: software
Abstract: CLBOS Reproducibility Package v2.2.0 This archive provides the software, experimental data, scripts, and derived results supporting the manuscript Admissible lower-bound orientation screening for modular agricultural coverage path planning, previously titled Certified Lower-Bound Orientation Search under Cross-Stage Objective Mismatch in Agricultural Coverage Path Planning. What's New in v2.2.0 Version 2.2.0 adds matched lower-bound timing evidence comparing the baseline admissible lower bound (LB0) with the MST-enhanced lower bound (LB1-MST). The experiment covers 18 preselected agricultural fields, five repetitions per method, and 180 formal execution records. The MST-enhanced lower bound reduced complete Route+Path evaluations from 355 to 225 (36.6%). The observed end-to-end runtime ratio (LB1/LB0) was 0.950, with a 95% field-cluster bootstrap confidence interval of [0.865, 1.073]. Since this interval includes 1, the experiment does not establish a statistically supported end-to-end speedup. The release includes the original execution records, environment fingerprints, analysis script, field-level summaries, independent verification script, and SHA-256 checksums. Previous Versions Version 2.1.0 added equal-scope runtime comparisons between CLBOS-MST and fixed Top-10 screening, together with cross-stage mismatch component decomposition and updated reproducibility mappings. Version 2.0.0 added transfer and sensitivity evaluations, controlled minimum-turning-radius analysis, field-difficulty analysis, repeated runtime measurements, and figure reproduction assets. All 120 files from the v2.1.0 archive are retained in v2.2.0, together with nine additional M2 evidence files. Reproduction Core numerical results: python3 scripts/reproduce_numbers.py Additional experimental analyses: python3 scripts/reproduce_v2_added.py Package integrity verification: python3 scripts/verify_package.py Independent verification of the new M2 experiment: python3 additional_evidence/m2_matched_bounds/CLBOS_M2_INDEPENDENT_VERIFY.py Scientific Scope and Limitations CLBOS provides instance-specific finite-domain stopping evidence under the stated planner semantics and admissibility assumptions. The reported optimality recovery results refer to the predefined discrete orientation domain and validated planner configuration. They do not constitute unconditional floating-point optimality certificates or establish performance guarantees across different coverage planners. The fixed Top-10 strategy remained exact in the tested transfer domains. CLBOS is not claimed to be uniformly faster than a successful fixed-budget screening strategy. Version 2.2.0 does not modify the CLBOS algorithm, underlying mathematical formulation, planner semantics, original 90-field benchmark, or previously reported experimental data. Version History and Identifiers v2.0.0: https://doi.org/10.5281/zenodo.21920938 v2.1.0: https://doi.org/10.5281/zenodo.21935272 Concept DOI: https://doi.org/10.5281/zenodo.21920937 Source repository: https://github.com/MFKD-Z/clbos-agricultural-cpp-reproducibility The v2.2.0 version-specific DOI is registered upon publication of the new Zenodo record.
Institution: Sichuan University
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
SDG: Zero hunger
Open Access status: green
OpenAlex ID: W7202363523