基于可达集和强化学习的智能汽车决策规划

高洪伟,尚秉旭,张鑫康,王洪峰,何维,裴晓飞

Decision-making and planning of intelligent vehicle based on reachable set and reinforcement learning

Hongwei GAO,Bingxu SHANG,Xinkang ZHANG,Hongfeng WANG,Wei HE,Xiaofei PEI

表 2 测试场景随机变量分布参数

Tab.2 Random variable distribution parameters in test scenario

参数 随机变量 1）注：U表示随机变量服从区间( )上的均匀分布. 车辆初始位置（前）/m $\begin{array}{*{20}{l}} {{{U}}(10,30)({\rm{checkpointA}})}^{1)}/ \\ {{{U}}(470,500)({\rm{checkpointC}})} \end{array}$ 车辆初始位置（中）/m $\begin{gathered} {{U}}(40,75)({\rm{checkpointA}})/ \\ {{U}}(510,545)({\rm{checkpointC}}) \\ \end{gathered} $ 车辆初始位置（后）/m $\begin{gathered} {{U}}(90,120)({\rm{checkpointA}})/ \\ {{U}}(560,590)({\rm{checkpointC}}) \\ \end{gathered} $ 静止车辆初始位置/m $ \begin{gathered} {{U(120,320)(}}{\rm{checkpointA}}{\text{)}}/ \\ {{U(590,760)(}}{\rm{checkpointC}}{\text{)}} \\ \end{gathered} $ 静止车辆压线概率/% $ 25$ 车辆初始车速/(m·s−1) $U(6,15)$