学会与环境自主互动的机器人
real-robots的Python项目详细描述
真正的机器人
学会与环境自主互动的机器人
安装
pip install -U real_robots
如果一切顺利，那么您应该能够运行：
real-robots-demo
它应该（最终）用一个小机械臂打开一扇小窗户 做一些随机的事情。
用法
importgymimportnumpyasnpimporttimeimportreal_robotsfromreal_robots.policyimportBasePolicyclassRandomPolicy(BasePolicy):def__init__(self,action_space):self.action_space=action_spaceself.action=np.zeros(action_space.shape[0])self.action+=-np.pi*0.5defstep(self,observation,reward,done):self.action+=0.4*np.pi*np.random.randn(self.action_space.shape[0])returnself.actionenv=gym.make("REALRobot-v0")pi=RandomPolicy(env.action_space)env.render("human")observation=env.reset()reward,done=0,Falsefortinrange(40):time.sleep(1./1000.)action=pi.step(observation,reward,done)observation,reward,done,info=env.step(action)print(t,reward)
本地评估
importgymimportnumpyasnpimportreal_robotsfromreal_robots.policyimportBasePolicyclassRandomPolicy(BasePolicy):def__init__(self,action_space):self.action_space=action_spaceself.action=np.zeros(action_space.shape[0])self.action+=-np.pi*0.5defstep(self,observation,reward,done):self.action+=0.4*np.pi*np.random.randn(self.action_space.shape[0])returnself.actionresult,detailed_scores=real_robots.evaluate(RandomPolicy,intrinsic_timesteps=40,extrinsic_timesteps=40,extrinsic_trials=5,visualize=True,goals_dataset_path="./goals.npy.npz",)# NOTE : You can find a sample goals.npy.npz file at## https://aicrowd-production.s3.eu-central-1.amazonaws.com/misc/REAL-Robots/goals.npy.npzprint(result)# {'score_total': 0.2327459533906755, 'score_2D': 0.6982378601720265, 'score_2.5D': 0, 'score_3D': 0}print(detailed_scores)# {'2D': [0.6646365565451159, 0.6632591441787807, 0.7569003923985664, 0.7167885964780916, 0.6896046112595778]}
另请参见我们的FAQ。
- 自由软件：麻省理工学院许可证
功能
- 待办事项
作者
- 弗朗西斯科·曼内拉
- 埃米利奥·卡通尼
- Sharada Mohanty
欢迎加入QQ群-->： 979659372
