forked from mpatacchiola/dissecting-reinforcement-learning
-
Notifications
You must be signed in to change notification settings - Fork 0
/
inverted_pendulum.py
170 lines (153 loc) · 7.64 KB
/
inverted_pendulum.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
#!/usr/bin/env python
# MIT License
# Copyright (c) 2017 Massimiliano Patacchiola
# https://mpatacchiola.github.io/blog/
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
# Implentation of the Inverted Pendulum problem using the notation of the book:
# 'Statistical Reinforcement Learning' by Masashi Sugiyama
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.animation as animation
class InvertedPendulum:
def __init__(self, pole_mass=2.0, cart_mass=8.0, pole_lenght=0.5, delta_t=0.1):
""" Create a new pendulum object.
It is possible to pass the parameter of the simulation.
@param pole_mass: the mass of the pole (default 2.0 Kg)
@param cart_mass: the mass of the cart (default 8.0 Kg)
@param pole_lenght: the lenght of the pole (default 0.5 m)
@param delta_t: the time step in seconds (default 0.1 s)
"""
self.angle_list = list()
self.gravity = 9.8
self.delta_t = delta_t
self.pole_mass = pole_mass
self.cart_mass = cart_mass
self.pole_lenght = pole_lenght
self.angle_t = np.random.normal(0, 0.05) # radians (vertical position)
self.angular_velocity_t = 0.0
self.alpha = 1.0 / (self.pole_mass + self.cart_mass)
def reset(self, exploring_starts=True, initial_angle=0.0):
""" It reset the pendulum to an initial position [0, 2*pi]
@param exploring_starts: if True a random position is taken
@param initial_angle: the initial position of the pendulum (requires exploring_starts=False)
@return: it returns the initial position of the pendulum and the velocity
"""
if exploring_starts:
initial_angle = np.random.uniform(0, np.pi/2.0)
#if initial_angle < -np.pi:
# initial_angle = -np.pi
#elif initial_angle > np.pi:
# initial_angle = np.pi
#else:
self.angle_t = np.random.normal(0, 0.1) # radians (vertical position)
self.angle_list = [] # clear the list
self.angular_velocity_t = 0.0
self.angle_list.append(initial_angle)
return [self.angle_t, self.angular_velocity_t]
def step(self, action):
"""Perform one step in the environment following the action.
@param action: an integer representing one of three actions [0, 1, 2]
where 0=move_left, 1=do_not_move, 2=move_right
@return: (angle_t1, angular_velocity_t1), reward, done
where reward is 0.0 when the pole is horizontal and 1.0 if vertical
done is True when the goal is reached
"""
if(action >= 3):
raise ValueError("[INVERTED PENDULUM][ERROR] The action value "
+ str(action) + " is out of range.")
done = False
reward = -0.01
action_list = [-50, 0, +50]
action_t = action_list[action]
angular_velocity_t1 = self.angular_velocity_t + \
(self.gravity * np.sin(self.angle_t) - \
self.alpha * self.pole_mass * self.pole_lenght * np.power(self.angular_velocity_t, 2) * (np.sin(2*self.angular_velocity_t)/2.0) + \
self.alpha * np.cos(self.angle_t) * action_t) / \
((4/3) * self.pole_lenght - self.alpha * self.pole_mass * self.pole_lenght * np.power(np.sin(self.angle_t), 2)) * self.delta_t
angle_t1 = self.angle_t + (angular_velocity_t1 * self.delta_t)
# Check the limit condition (horizontal pole)
if angle_t1 < -(np.pi/2.0):
angle_t1 = -(np.pi/2.0)
angular_velocity_t1 = 0
if angle_t1 > (np.pi/2.0):
angle_t1 = (np.pi/2.0)
angular_velocity_t1 = 0
# Assign the new position and velocity
self.angle_t = angle_t1
self.angular_velocity_t= angular_velocity_t1
self.angle_list.append(angle_t1)
# Reward and done
if angle_t1 >= (np.pi/2.0) or angle_t1 <= -(np.pi/2.0):
reward = 0.0
done = True
else:
reward = np.cos(angle_t1)
done = False
# Return state_t1, reward, done
return [angle_t1, angular_velocity_t1], reward, done
def render(self, file_path='./inverted_pendulum.mp4', mode='mp4'):
""" When the method is called it saves an animation
of the steps happened until that point in the episode.
Ideally it should be called at the end of the episode,
or every k episodes.
ATTENTION: It requires avconv and/or imagemagick installed.
@param file_path: the name and path of the video file
@param mode: the file can be saved as 'gif' or 'mp4'
"""
# Plot init
fig = plt.figure()
axis_limit = self.pole_lenght + (self.pole_lenght * 0.5)
ax = fig.add_subplot(111, autoscale_on=False, xlim=(-axis_limit, axis_limit), ylim=(0.0, 1.5*axis_limit))
ax.grid(False) # disable the grid
ax.set_aspect('equal')
ax.set_yticklabels([])
# x_line = np.linspace(start=-axis_limit, stop=axis_limit, num=100)
# y_line = np.zeros(100)
# ax.plot(x_line, y_line) # plot the base-line
# line, _ = ax.plot(x, y, 'o-', lw=2)
line, = ax.plot([], [],color='black', linestyle='solid', linewidth=1.5, marker='o', markerfacecolor='#aa0000', markersize=10, zorder=1)
# Adding the brown circle pad
circle = plt.Circle((0.0,-0.01), radius=0.05, color='#2b2200', fill=True, zorder=2)
ax.add_patch(circle)
# Adding the text
time_text = ax.text(0.05, 0.85, '', transform=ax.transAxes)
_angle_list = self.angle_list
_delta_t = self.delta_t
def _init():
line.set_data([], [])
time_text.set_text('')
return line, time_text
def _animate(i):
angle_cos = np.cos(_angle_list[i]) * self.pole_lenght
angle_sin = np.sin(_angle_list[i]) * self.pole_lenght
x1, y1 = [0, angle_sin], [0, angle_cos]
#y1 = (angle_cos, angle_sin)
line.set_data(x1, y1)
time_text.set_text("Time: " + str(np.round(i*_delta_t, 1)) + "s" + '\n' + "Frame: " + str(i))
return line, time_text
ani = animation.FuncAnimation(fig, _animate, np.arange(1, len(self.angle_list)),
blit=True, init_func=_init, repeat=False)
if mode == 'gif':
ani.save(file_path, writer='imagemagick', fps=int(1/self.delta_t))
elif mode == 'mp4':
ani.save(file_path, fps=int(1/self.delta_t), writer='avconv', codec='libx264')
# Clear the figure
fig.clear()
plt.close(fig)