if __name__ == '__main__':
max_steps = 100
max_iters = 1000
seed = random.randint(0, 100)
agent = qAgent()
grid = GridWorld(size=8, force_fast=True, seed=seed)
grid.show()
print()
for iter in range(max_iters):
agent.set_grid(grid)
i, j = 0, 0 # initial state
cum_reward = 0
for step in range(max_steps):
action = agent.get_action(i, j)
new_i, new_j = grid.move(i, j, action)
reward, is_final = grid.get_reward(i, j)
cum_reward += reward
agent.update_q(i, j, new_i, new_j, action, reward)
if is_final:
break
i = new_i
j = new_j
if iter % 100 == 0:
print(
'Episode {} finished after {} steps with cumulative reward of {}'
.format(iter, step, cum_reward))
grid = GridWorld(size=8, force_fast=True, seed=seed)
print()
show_qtable(agent, grid.size)
