tcship

python package for tcship race.

INSTALL

pip install tcship

SAMPLE

default with display

from tcShip import tc_ship,config
import random

def step_model(env, observation):
    ship = observation["body"][0]
    t1 = observation["target"][0]
    t2 = observation["target"][1]
    stone = observation['stone']

    # 曼哈顿距离
    d1 = abs(t1[0] - ship[0]) + abs(t1[1] - ship[1])
    d2 = abs(t2[0] - ship[0]) + abs(t2[1] - ship[1])
    if d1 > d2:
        i = 1
    else:
        i = 0
    # 最近星球坐标
    target = observation["target"][i]

    if ship[0] > target[0]:
        next_step = "left"
        next_ship = [ship[0] - 1, ship[1]]
    elif ship[0] < target[0]:
        next_step = "right"
        next_ship = [ship[0] + 1, ship[1]]
    else:
        next_step = "down"
        next_ship = [ship[0], ship[1] + 1]
    # 思考下什么时候向上或者向下？
    # elif ship[1] > target[1]:
    #     next_step =  "up"
    #     next_ship = [ship[0], ship[1] - 1]

    #检查下下一步会不会撞上陨石
    if next_ship not in stone:
        return next_step
    #仿照检测陨石，如何实现增加检测边界横坐标和纵坐标都不能 < 0或 > 16 ？#and ?>=0 and ?<= 16
    else:
        safe_action = ["right","left","up","down"]
        safe_action.remove(next_step)
        #思考是否有更好的方向选择替代随机选择？让飞船更快走到行星,以及此时随机选择出来的方向就一定安全么？
        #比如同时可以选的方向为"up"和"down"，如果行星在飞船上方，那我们优先选择"up"。
        return random.choice(safe_action)
        ##思考下飞船方向能不能直接回头？如何避免？
        # global last_action
        # if last_action == "left" and action == "right":

def main():
    env = tc_ship()
    observation = env.reset()
    rew = 0
    for _ in range(100):
        action = step_model(env, observation)
        print(action)
        observation, reward, done, info = env.step(action)
        rew += reward
        env.sleep(0.5)

        if done:
            print("rew :  " + str(rew))
            break

main()

config no display

from tcShip import tc_ship, config
import random

config.WORKSPACE = "tianchi" #"local"则展示可视化动画
#正式版代码模版 
def step_model(env, observation):
    action_space = env.action_space
    action = random.choice(action_space)
    return action

def main():
    env = tc_ship()
    observation = env.reset()
    rew = 0
    for _ in range(100):
        action = step_model(env, observation)
        print(action)
        observation, reward, done, info = env.step(action)
        rew += reward
        env.sleep(0.5)

        if done:
            print("rew :  " + str(rew))
            break

main()