An Assignment-Based Approach to Efficient Real-Time City-Scale Taxi Dispatching

摘要: This study proposes and evaluates an efficient real-time taxi dispatching strategy that solves the linear assignment problem to find a globally optimal taxi-to-request at each decision epoch. The authors compare assignment-based with two popular rule-based strategies. They evaluate strategies in detail city of Berlin neighboring region Brandenburg using microscopic large-scale MATSim simulator. produced better results for both drivers (less idle driving) passengers waiting). However, computing assignments thousands taxis huge road network turned out be computationally demanding. Certain adaptations pertaining cost matrix calculation were necessary increase computational efficiency assure responsiveness.