Courier Route Optimization Using Google OR-Tools on Asymmetric Road Networks

The last-mile delivery phase represents the most critical segment of modern logistics supply chains, frequently suffering from operational cost and time inefficiencies. In dense urban residential areas, couriers typically determine delivery sequences manually based on the intuitive nearest-location heuristic (Nearest Neighbor). This manual approach frequently leads to inefficient backtracking routes due to asymmetric real-world road networks, such as one-way regulations and narrow residential alleys. This study implements the Google Operations Research Tools (OR-Tools) library to optimize courier delivery routes based on real road network topology. Destination coordinates were extracted from OpenStreetMap geospatial data in the Telaga Murni residential area, Cikarang Barat, and the actual driving distance matrix was computed using the Open Source Routing Machine (OSRM). The routing model was solved using a two-phase architecture: initial route construction (Path Cheapest Arc) and route refinement (Guided Local Search). The evaluation was conducted across three distinct delivery scales: 10 packages (small), 30 packages (medium), and 50 packages (large). Experimental results demonstrate that Google OR-Tools consistently outperforms manual courier routing, achieving distance reductions of 13.91% for 10 packages (from 3.33 km to 2.86 km), 19.97% for 30 packages (from 5.64 km to 4.50 km), and 19.97% for 50 packages (from 7.30 km to 5.84 km). The initial heuristic route construction was completed in under 0.5 seconds. The integration of Google OR-Tools and OSRM provides a dependable, lightweight, and practical routing solution for reducing courier operational mileage at zero software licensing cost.