OpenAI Gym CartPole - Q table (get state)

q_table_get_state.py

def get_state(observation, n_buckets, state_bounds):
    state = [0] * len(observation) 
    for i, s in enumerate(observation): # 每個 feature 有不同的分配
        l, u = state_bounds[i][0], state_bounds[i][1] # 每個 feature 值的範圍上下限
        if s <= l: # 低於下限，分配為 0
            state[i] = 0
        elif s >= u: # 高於上限，分配為最大值
            state[i] = n_buckets[i] - 1
        else: # 範圍內，依比例分配
            state[i] = int(((s - l) / (u - l)) * n_buckets[i])

    return tuple(state)