Q_Learning_step1.m

function [state,action,r,flag_break,speed_new] = Q_Learning_step1(Q,LINK,TLINK,IndexOfCell,numOfCell,maxSpeed,p,epsilon,nominalSpeed,lane_id)
% Actionspace index
% 1:a=3
% 2:a=1
% 3:a=0
% 4:a=-1
% 5:a=-3
% 6:换道，a=0
a_0=[3,1,0,-1,-3,0]; %加速度
flag_break=false;

% get state and actionspace
[state,ActionSpace] = GetStateAction(LINK,TLINK,IndexOfCell,numOfCell,maxSpeed,p,lane_id);

% choose action（acc） according to epsilon-greedy policy
if rand() <= epsilon  %epsilon=0.1
    action = ActionSpace(uint16(rand*(length(ActionSpace)-1)+1)); 
else
    [~, action] = max(Q(state,ActionSpace));
end

%速度更新
if action == 6
    flag_break = true; %如果编号为6，则需变道，需要在主函数中跳出当层循环
    speed_new = LINK(IndexOfCell);
    
else
    a=a_0(action); %得到加速度
    speed_new = LINK(IndexOfCell)+a;
    speed_new = max(0,speed_new);
    emptyFront = GetEmptyFront(LINK, numOfCell, maxSpeed, IndexOfCell);
    speed_new = min(speed_new,emptyFront);
    %随机慢化
    r=unifrnd(0,1);
    if r<=0.05 %随机慢化概率
       speed_new=speed_new-1; 
    end
    speed_new= max(0,speed_new);%保证不倒车
end

%定义奖励函数
%定义加速度惩罚
if action ==2 || action==5  %强加或强减速，惩罚-5
    acc=-5;
elseif action==3 || action==6 %保持，不惩罚
    acc=0;
else %其余，惩罚-1
    acc=-1;
end
% r = GetReward(nominalSpeed,speed_new);    
r = GetReward(nominalSpeed,speed_new,acc);

end