This repository has been archived by the owner on May 13, 2021. It is now read-only.
forked from EivMeyer/gym-auv
-
Notifications
You must be signed in to change notification settings - Fork 3
/
utils.py
231 lines (198 loc) · 9.27 KB
/
utils.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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
import argparse
import os
import matplotlib.pyplot as plt
import numpy as np
from pandas import DataFrame
from cycler import cycler
from gym_auv.utils.controllers import PI, PID
PI = PI()
PID_cross = PID(Kp=1.8, Ki=0.01, Kd=0.035)
PID_cross = PID(Kp=1.8, Ki=0.01, Kd=0.035)
def parse_experiment_info():
"""Parser for the flags that can be passed with the run/train/test scripts."""
parser = argparse.ArgumentParser()
parser.add_argument("--exp_id", type=int, help="Which experiment number to run/train/test")
parser.add_argument("--scenario", default="expert", type=str, help="Which scenario to run")
parser.add_argument("--controller_scenario", default=None, type=str, help="Which scenario the agent was trained in")
parser.add_argument("--controller", default=None, type=int, help="Which model to load as main controller. Requires only integer")
args = parser.parse_args()
experiment_dir = os.path.join(r"./log", r"Experiment {}".format(args.exp_id))
if args.controller_scenario is not None:
agent_path = os.path.join(experiment_dir, args.controller_scenario, "agents")
else:
agent_path = os.path.join(experiment_dir, args.scenario, "agents")
if args.controller is not None:
agent_path = os.path.join(agent_path, "model_" + str(args.controller) + ".pkl")
else:
agent_path = os.path.join(agent_path, "best_model.pkl")
return experiment_dir, agent_path, args.scenario
def calculate_IAE(sim_df):
"""
Calculates and prints the integral absolute error provided an environment id and simulation data
"""
IAE_cross = sim_df[r"e"].abs().sum()
IAE_vertical = sim_df[r"h"].abs().sum()
print("IAE Cross track: {}, IAE Vertical track: {}".format(IAE_cross, IAE_vertical))
return IAE_cross, IAE_vertical
def simulate_environment(env, agent):
global error_labels, current_labels, input_labels, state_labels
state_labels = [r"$N$", r"$E$", r"$D$", r"$\phi$", r"$\theta$", r"$\psi$", r"$u$", r"$v$", r"$w$", r"$p$", r"$q$", r"$r$"]
current_labels = [r"$u_c$", r"$v_c$", r"$w_c$"]
input_labels = [r"$\eta$", r"$\delta_r$", r"$\delta_s$"]
error_labels = [r"$\tilde{u}$", r"$\tilde{\chi}$", r"e", r"$\tilde{\upsilon}$", r"h"]
labels = np.hstack(["Time", state_labels, input_labels, error_labels, current_labels])
done = False
env.reset()
while not done:
action = agent.predict(env.observation, deterministic=True)[0]
_, _, done, _ = env.step(action)
errors = np.array(env.past_errors)
time = np.array(env.time).reshape((env.total_t_steps,1))
sim_data = np.hstack([time, env.past_states, env.past_actions, errors, env.current_history])
df = DataFrame(sim_data, columns=labels)
error_labels = [r"e", r"h"]
return df
def set_default_plot_rc():
"""Sets the style for the plots report-ready"""
colors = (cycler(color= ['#EE6666', '#3388BB', '#88DD89', '#EECC55', '#88BB44', '#FFBBBB']) +
cycler(linestyle=['-', '-', '-', '--', ':', '-.']))
plt.rc('axes', facecolor='#ffffff', edgecolor='black',
axisbelow=True, grid=True, prop_cycle=colors)
plt.rc('grid', color='gray', linestyle='--')
plt.rc('xtick', direction='out', color='black', labelsize=14)
plt.rc('ytick', direction='out', color='black', labelsize=14)
plt.rc('patch', edgecolor='#ffffff')
plt.rc('lines', linewidth=4)
def plot_attitude(sim_df):
"""Plots the state trajectories for the simulation data"""
set_default_plot_rc()
ax = sim_df.plot(x="Time", y=[r"$\phi$",r"$\theta$", r"$\psi$"], kind="line")
ax.set_xlabel(xlabel="Time [s]",fontsize=14)
ax.set_ylabel(ylabel="Angular position [rad]",fontsize=14)
ax.legend(loc="lower right", fontsize=14)
ax.set_ylim([-np.pi,np.pi])
plt.show()
def plot_velocity(sim_df):
"""Plots the velocity trajectories for the simulation data"""
set_default_plot_rc()
ax = sim_df.plot(x="Time", y=[r"$u$",r"$v$"], kind="line")
ax.plot(sim_df["Time"], sim_df[r"$w$"], dashes=[3,3], color="#88DD89", label=r"$w$")
ax.plot([0,sim_df["Time"].iloc[-1]], [1.5,1.5], label=r"$u_d$")
ax.set_xlabel(xlabel="Time [s]", fontsize=14)
ax.set_ylabel(ylabel="Velocity [m/s]", fontsize=14)
ax.legend(loc="lower right", fontsize=14)
ax.set_ylim([-0.25,2.25])
plt.show()
def plot_angular_velocity(sim_df):
"""Plots the angular velocity trajectories for the simulation data"""
set_default_plot_rc()
ax = sim_df.plot(x="Time", y=[r"$p$",r"$q$", r"$r$"], kind="line")
ax.set_xlabel(xlabel="Time [s]", fontsize=14)
ax.set_ylabel(ylabel="Angular Velocity [rad/s]", fontsize=14)
ax.legend(loc="lower right", fontsize=14)
ax.set_ylim([-1,1])
plt.show()
def plot_control_inputs(sim_dfs):
""" Plot control inputs from simulation data"""
set_default_plot_rc()
c = ['#EE6666', '#88BB44', '#EECC55']
for i, sim_df in enumerate(sim_dfs):
control = np.sqrt(sim_df[r"$\delta_r$"]**2+sim_df[r"$\delta_s$"]**2)
plt.plot(sim_df["Time"], sim_df[r"$\delta_s$"], linewidth=4, color=c[i])
plt.xlabel(xlabel="Time [s]", fontsize=14)
plt.ylabel(ylabel="Normalized Input", fontsize=14)
plt.legend(loc="lower right", fontsize=14)
plt.legend([r"$\lambda_r=0.9$", r"$\lambda_r=0.5$", r"$\lambda_r=0.1$"], loc="upper right", fontsize=14)
plt.ylim([-1.25,1.25])
plt.show()
def plot_control_errors(sim_dfs):
"""
Plot control inputs from simulation data
"""
#error_labels = [r'e', r'h']
set_default_plot_rc()
c = ['#EE6666', '#88BB44', '#EECC55']
for i, sim_df in enumerate(sim_dfs):
error = np.sqrt(sim_df[r"e"]**2+sim_df[r"h"]**2)
plt.plot(sim_df["Time"], error, linewidth=4, color=c[i])
plt.xlabel(xlabel="Time [s]", fontsize=12)
plt.ylabel(ylabel="Tracking Error [m]", fontsize=12)
#plt.ylim([0,15])
plt.legend([r"$\lambda_r=0.9$", r"$\lambda_r=0.5$", r"$\lambda_r=0.1$"], loc="upper right", fontsize=14)
plt.show()
def plot_3d(env, sim_df):
"""
Plots the AUV path in 3D inside the environment provided.
"""
plt.rcdefaults()
plt.rc('lines', linewidth=3)
ax = env.plot3D()#(wps_on=False)
ax.plot3D(sim_df[r"$N$"], sim_df[r"$E$"], sim_df[r"$D$"], color="#EECC55", label="AUV Path")#, linestyle="dashed")
ax.set_xlabel(xlabel="North [m]", fontsize=14)
ax.set_ylabel(ylabel="East [m]", fontsize=14)
ax.set_zlabel(zlabel="Down [m]", fontsize=14)
ax.legend(loc="upper right", fontsize=14)
plt.show()
def plot_multiple_3d(env, sim_dfs):
"""
Plots multiple AUV paths in 3D inside the environment provided.
"""
plt.rcdefaults()
c = ['#EE6666', '#88BB44', '#EECC55']
styles = ["dashed", "dashed", "dashed"]
plt.rc('lines', linewidth=3)
ax = env.plot3D()#(wps_on=False)
for i,sim_df in enumerate(sim_dfs):
ax.plot3D(sim_df[r"$N$"], sim_df[r"$E$"], sim_df[r"$D$"], color=c[i], linestyle=styles[i])
ax.set_xlabel(xlabel="North [m]", fontsize=14)
ax.set_ylabel(ylabel="East [m]", fontsize=14)
ax.set_zlabel(zlabel="Down [m]", fontsize=14)
ax.legend(["Path",r"$\lambda_r=0.9$", r"$\lambda_r=0.5$",r"$\lambda_r=0.1$"], loc="upper right", fontsize=14)
plt.show()
def plot_current_data(sim_df):
set_default_plot_rc()
#---------------Plot current intensity------------------------------------
ax1 = sim_df.plot(x="Time", y=current_labels, linewidth=4, style=["-", "-", "-"] )
ax1.set_title("Current", fontsize=18)
ax1.set_xlabel(xlabel="Time [s]", fontsize=14)
ax1.set_ylabel(ylabel="Velocity [m/s]", fontsize=14)
ax1.set_ylim([-1.25,1.25])
ax1.legend(loc="right", fontsize=14)
#ax1.grid(color='k', linestyle='-', linewidth=0.1)
plt.show()
#---------------Plot current direction------------------------------------
"""
ax2 = ax1.twinx()
ax2 = sim_df.plot(x="Time", y=[r"$\alpha_c$", r"$\beta_c$"], linewidth=4, style=["-", "--"] )
ax2.set_title("Current", fontsize=18)
ax2.set_xlabel(xlabel="Time [s]", fontsize=12)
ax2.set_ylabel(ylabel="Direction [rad]", fontsize=12)
ax2.set_ylim([-np.pi, np.pi])
ax2.legend(loc="right", fontsize=12)
ax2.grid(color='k', linestyle='-', linewidth=0.1)
plt.show()
"""
def plot_collision_reward_function():
horizontal_angles = np.linspace(-70, 70, 300)
vertical_angles = np.linspace(-70, 70, 300)
gamma_x = 25
epsilon = 0.05
sensor_readings = 0.4*np.ones((300,300))
image = np.zeros((len(vertical_angles), len(horizontal_angles)))
for i, horizontal_angle in enumerate(horizontal_angles):
horizontal_factor = (1-(abs(horizontal_angle)/horizontal_angles[-1]))
for j, vertical_angle in enumerate(vertical_angles):
vertical_factor = (1-(abs(vertical_angle)/vertical_angles[-1]))
beta = horizontal_factor*vertical_factor + epsilon
image[j,i] = beta*(1/(gamma_x*(sensor_readings[j,i])**4))
print(image.round(2))
ax = plt.axes()
plt.colorbar(plt.imshow(image),ax=ax)
ax.imshow(image, extent=[-70,70,-70,70])
ax.set_ylabel("Vertical vessel-relative sensor angle [deg]", fontsize=14)
ax.set_xlabel("Horizontal vessel-relative sensor angle [deg]", fontsize=14)
ax.xaxis.set_tick_params(labelsize=14)
ax.yaxis.set_tick_params(labelsize=14)
plt.show()
if __name__ == "__main__":
plot_collision_reward_function()