摘要
出租车运载体系主要存在以下问题:出租车空载率高。资源调用不合理,导致整个出租车行业形象受损,恶性循环,使得出租车行业整体的发展受到阻碍。为了解决以上问题,文章将时间差分方法引入了出租车智能规划算法中,利用SARSA和Q-learning的方法,使智能体在模拟地图中不断寻找路线,通过和环境进行交互来选择当前状态的动作,增加对环境的熟悉程度并且不断更新路线策略。避免了传统方法中收敛过早导致准确性不足的情况。通过模型仿真证明将时间差分法应用在基础的路径规划中在模拟环境中是可行的,能够合理避开障碍物和一些模拟的需要规避的禁行区域。
There are mainly the following problems in the taxi delivery system:high no load rate of taxis.Unreasonable resource allocation leads to the damage to the image of the entire taxi industry and a vicious circle,which hinders the development of the taxi industry as a whole.In order to solve the above problems,the time difference method is introduced into the taxi intelligent planning algorithm.Using the methods of SARSA and Q-learning,the agent can constantly find the route in the simulated map,select the action of the current state by interacting with the environment,increase the familiarity with the environment and constantly update the route strategy.It avoids the situation that the accuracy is not high enough due to the premature convergence in the traditional methods.Through the model simulation,it is proved that the application of the time difference method in the basic path planning is feasible in the simulation environment,and it can reasonably avoid obstacles as well as some simulated forbidden areas that need to be avoided.
出处
《科技创新与应用》
2021年第1期26-29,共4页
Technology Innovation and Application
