无人驾驶汽车共享调度方法研究

Study on Scheduling Method of Shared Driverless Vehicles

由于无人驾驶汽车不存在人为驾驶的随机性,操控精确性高,组成的交通流通行能力大,合理的调度方法能够提高无人驾驶汽车调度系统的整体收益、降低服务时间、降低乘客的出行费用,并从根本上缓解城市交通安全、交通拥堵、交通污染等社会问题。为使无人驾驶汽车更好地为乘客提供共享乘车服务,在深入了解无人驾驶汽车技术发展的基础上结合共享调度理论,建立了适合于一定区域内满足乘客需求的共享无人驾驶汽车调度模型。对模型中的公式以及限定条件进行了严密的分析,根据模型的限制条件设计了一种新型的乘客合乘的算法,新提出的算法首先研究了如何确定请求-路线-车辆分配情况,动态生成在线需求和无人驾驶车辆的最佳路线,然后利用贪婪分配进行匹配,最后完善已有算法的不足,通过优化约束条件来改进匹配,通过多次迭代后得到最优分配。为了更好地验证新提出算法的运算效率,利用某地区的OD调查数据对新提出的算法和聚类算法从匹配成功率、算法平均运行时间、乘客平均等待时间3个角度应用软件进行模拟分析,表明新提出的算法运算效率得到较大提高。

Due to the absence of the randomness of human driving in driverless vehicle, high control accuracy and large traffic flow capacity, a reasonable dispatching method can improve the overall revenue of the dispatching system of driverless vehicle, reduce service time and passenger travel costs, and fundamentally alleviate the social problems such as urban traffic safety traffic congestion, and traffic pollution. In order to make driverless vehicles better provide shared ridership service for passengers, on the basis of in-depth understanding the development of driverless vehicle technology, combining with shared scheduling theory, a shared driverless vehicle scheduling model to meet the needs of passengers in a certain area is established. The formulas and constraints in the model are analyzed rigorously. According to the constraints of the model, a new car-sharing algorithm for passengers is designed. First, the way of determining the request-route-vehicle assignment is studied in the new algorithm, the online demand and the optimal route for driverless vehicles are dynamically generated. Then, the greedy allocation is used for matching. Finally, the shortcomings of existing algorithms are improved, the constraint conditions are optimized to improve the matching, and the optimal allocation is obtained after several iterations. In order to better verify the efficiency of the new algorithm, by using the OD survey data of a certain area, the proposed algorithm and the clustering algorithm are simulated and analyzed from the aspects of matching success rate, average running time of algorithm and average waiting time of passengers. The results show that the efficiency of the new algorithm is greatly improved.