q-learning算法路径规划matlab代码

以下是一个简单的Q-learning算法的路径规划Matlab代码示例：

% 初始状态
start_state = 1;

% 终止状态
goal_state = 6;

% 状态转移矩阵
R = [0 -1 -1 0 0 0;
     0 0 -1 -1 0 -1;
     0 -1 0 -1 -1 -1;
     0 -1 -1 0 -1 -1;
     -1 0 -1 -1 0 -1;
     -1 -1 0 0 -1 0];

% 转移概率矩阵
T = [0.5 0.5 0 0 0 0;
     0 0.5 0.5 0 0 0;
     0 0 0.5 0.5 0 0;
     0 0 0 0.5 0.5 0;
     0 0 0 0 0.5 0.5;
     0 0 0 0 0 1];

% Q-table初始化
Q = zeros(size(T));

% 超参数设定
learning_rate = 0.8;
discount_factor = 0.95;
num_episodes = 1000;

% Q-learning算法
for episode = 1:num_episodes
    % 初始化当前状态
    current_state = start_state;
    while current_state ~= goal_state
        % 根据epsilon-greedy策略选择动作
        if rand < 0.1
            action = randi([1 size(T, 2)]);
        else
            [~, action] = max(Q(current_state, :));
        end
        
        % 执行动作并转移到下一个状态
        next_state = randsample(1:size(T, 1), 1, true, T(current_state, :));
        
        % 更新Q-table
        Q(current_state, action) = Q(current_state, action) + learning_rate * (R(current_state, next_state) + discount_factor * max(Q(next_state, :)) - Q(current_state, action));
        
        % 更新当前状态
        current_state = next_state;
    end
end

% 输出最优路径
path = start_state;
current_state = start_state;
while current_state ~= goal_state
    [~, action] = max(Q(current_state, :));
    current_state = action;
    path = [path current_state];
end

disp(path);

该代码中，我们使用一个6状态的示例环境，其中0表示无效状态，-1表示可行状态，1表示目标状态。状态转移矩阵R和转移概率矩阵T也在代码中定义。在Q-table初始化后，我们使用Q-learning算法进行训练。最后，我们输出从起始状态到目标状态的最优路径。

请注意，此代码仅作为示例，您需要根据您的具体问题进行修改。