Intelligent Truck Matching in Full Truckload Shipments using Ping2Hex approach

Project44 deployed Intelligent Truck Matching 2.0, a machine learning system that uses Uber H3 hexagonal spatial indexing and LightGBM gradient boosting to match trucks with shipments when GPS data is incomplete or corrupted. The system achieves 26 percentage point precision improvements in North America and doubles coverage, addressing a critical supply chain visibility challenge.

Project44's deployment of ITM 2.0 represents a meaningful advance in supply chain logistics technology by solving a practical problem that traditional rule-based systems cannot handle. When vehicle identifiers go missing or corrupt—a common occurrence in real-world GPS data—shipments lose visibility entirely. By reformulating truck matching as a probabilistic ranking problem rather than a deterministic lookup, ITM 2.0 enables continued tracking even under degraded data conditions.

The technical approach demonstrates sophisticated feature engineering: discretizing GPS pings into hexagonal spatial indices captures route geometry independent of exact coordinates, while temporal features add context about movement patterns. LightGBM's gradient boosting handles the probabilistic ranking efficiently, and threshold-based post-processing provides interpretability for production deployment. The system's robustness to 1 km geocoding errors and sparse ping frequencies shows practical engineering for real-world conditions rather than idealized scenarios.

For logistics operators and freight visibility platforms, this development has meaningful business implications. Better truck-to-shipment matching directly improves ETA prediction accuracy, reducing shipper uncertainty and enabling more efficient supply chain coordination. The 26 percentage point precision gain in North America represents substantial operational value. For Project44's customers and the broader logistics-as-a-platform ecosystem, this type of AI-driven matching becomes a competitive differentiator in an industry where visibility directly correlates with customer satisfaction and retention.

The production deployment at scale across full truckload operations validates the approach's viability. Future development likely focuses on expanding geographic coverage, reducing false positive matches, and potentially extending the methodology to less-than-truckload shipments where matching complexity increases significantly.

• →ITM 2.0 uses hexagonal spatial indexing and LightGBM to match trucks with shipments despite missing or corrupted GPS identifiers.
• →The system achieves 26 percentage point precision improvement in North America while doubling shipment visibility coverage.
• →Machine learning probabilistic ranking outperforms traditional rule-based matching when vehicle data is incomplete or degraded.
• →Production deployment demonstrates robustness to real-world challenges including 1 km geocoding errors and sparse GPS pings.
• →Improved truck-to-shipment matching directly enhances ETA prediction accuracy and supply chain visibility for logistics operators.