Online Taxi Dispatching Algorithm Based on Quantum Annealing

Abstract

Quantum annealing techniques are mainly used to solve optimization and sampling problems, which have a greater potential to obtain globally optimal solutions for specific combinatorial optimization problems due to their unique quantum tunneling properties. Firstly, this paper transforms the taxi repositioning problem into a combinatorial optimization problem that can be represented by a QUBO model with exponentially growing path solutions. The optimal solution aims at repositioning the online taxis to optimize the passenger waiting time and the average daily revenue of taxi drivers. Secondly, this paper proposes a QUBO formulation to solve a specific taxi repositioning problem, which optimizes the traffic distribution of taxis based on the existing spatial transfer probability to obtain the unique target region selection result. Finally, the simulation results on the real trajectory data of taxis in Chengdu in November 2016 show that the solution is nearly optimal, and the quantum annealing algorithm optimizes about 7% of the indicator than the simulated annealing algorithm by making each vehicle select the unique region to accurately reduce the passenger waiting time while improving the driver's revenue.