-
Notifications
You must be signed in to change notification settings - Fork 0
/
vis_repertoire.py
213 lines (164 loc) · 7.85 KB
/
vis_repertoire.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
import sys, os
import argparse
import pandas as pd
import numpy as np
from mpl_toolkits.mplot3d import Axes3D
import matplotlib
import matplotlib.pyplot as plt
import src.torch.pytorch_util as ptu
#from src.envs.hexapod_dart.hexapod_env_v2 import simulate
from src.envs.hexapod_dart.hexapod_env import HexapodEnv
from src.models.dynamics_models.deterministic_model import DeterministicDynModel
dynamics_model = DeterministicDynModel(48,18,500)
dynamics_model = ptu.load_model(dynamics_model, "hexapod_detdyn_archcond_trained.pth")
env = HexapodEnv(dynamics_model=dynamics_model, render=True, ctrl_freq=100)
from src.envs.panda_bullet.pushing_env import pandaPushingEnv
env = pandaPushingEnv(dynamics_model, gui=True)
def key_event(event, args):
if event.key == 'escape':
sys.exit(0)
def click_event(event, args):
'''
# reutrns a list of tupples of x-y points
click_in = plt.ginput(1,-1) # one click only, should not timeout
print("click_in: ",click_in)
selected_cell = [int(click_in[0][0]), int(click_in[0][1])]
print(selected_cell)
selected_x = selected_cell[0]
selected_y = selected_cell[1]
'''
#event.button ==1 is the left mouse click
if event.button == 1:
selected_x = int(event.xdata)
selected_y = int(event.ydata)
selected_solution = data[(data["x_bin"] == selected_x) & (data["y_bin"] == selected_y)]
#selected_solution = data[(data["y_bin"] == selected_x) & (data["z_bin"] == selected_y)]
# For hexapod omnitask
#selected_ctrl = selected_solution.iloc[:,3:39].to_numpy()
#print(selected_ctrl[0].shape) #(1,36)
#hexapod uni
#selected_solution = selected_solution.iloc[0, :]
#selected_ctrl = selected_solution.iloc[8:-2].to_numpy()
# For panda bullet
#print("Selected_solution: ", selected_solution.iloc[0,:])
selected_solution = selected_solution.iloc[0,:]
selected_ctrl = selected_solution.iloc[4:-4].to_numpy()
print("Selected ctrl shape: ", selected_ctrl.shape) # should be 3661
print("Selected descriptor bin: " ,selected_x, selected_y)
print("Selected descriptor from archive: ", selected_solution.iloc[1:7].to_numpy())
print("Selected fitness from archive: ", selected_solution.iloc[0])
# ---- SIMULATE THE SELECTED CONTROLLER -----#
#simulate(selected_ctrl, 5.0, render=True) # Hexapod
env.evaluate_solution(selected_ctrl)
'''
fit, desc, _, _ = env.evaluate_solution_uni(selected_ctrl, render=True)
print("fitness from simulation real eval:", fit)
fit, desc, _, _ = env.evaluate_solution_model_uni(selected_ctrl)
print("fitness from dynamics model :", fit)
'''
#env.evaluate_solution(selected_ctrl, render=True)
#simulate(selected_ctrl, render=True) # panda bullet
#simulate(selected_ctrl, 5.0, render=True) # panda dart
#evaluate_solution(selected_ctrl, gui=True)
print("SIMULATION DONE")
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--filename", type=str) # file to visualize rollouts from
parser.add_argument("--sim_time", type=float, help="simulation time depending on the type of archive you chose to visualize, 3s archive or a 5s archive")
parser.add_argument("--plot_type", type=str, default="scatter", help="scatter plot or grid plot")
args = parser.parse_args()
data = pd.read_csv(args.filename)
data = data.iloc[:,:-1] # drop the last column which was made because there is a comma after last value i a line
#data = np.loadtxt(args.filename)
#fit_bd = data.iloc[:,0:3]
#print("ARCHIVE SHAPE ORIGINAL: ", data.shape)
# Get names of indexes for which column Age has value 30
#indexNames = data[data.iloc[:,0] <= -0.1].index
# Delete these row indexes from dataFrame
#data.drop(indexNames, inplace=True)
print("ARCHIVE SHAPE AFTER DROPPING: ", data.shape)
#print(data)
#print(fit_bd)
'''
#For Hexapod
data['x_bin']=pd.cut(x = data.iloc[:,4],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
data['y_bin']=pd.cut(x = data.iloc[:,5],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
'''
'''
# hexapod uni
data['bd1_bin']=pd.cut(x = data.iloc[:,1],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
data['bd2_bin']=pd.cut(x = data.iloc[:,2],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
` data['bd3_bin']=pd.cut(x = data.iloc[:,3],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
data['bd4_bin']=pd.cut(x = data.iloc[:,4],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
data['bd5_bin']=pd.cut(x = data.iloc[:,5],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
data['bd6_bin']=pd.cut(x = data.iloc[:,6],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
'''
#For panda push
data['scaled_x'] = (data.iloc[:,1]+0.5)
data['scaled_y'] = (data.iloc[:,2]+0.5)
data['x_bin']=pd.cut(x = data['scaled_x'],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
data['y_bin']=pd.cut(x = data['scaled_y'],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
'''
data['scaled_x'] = ((data.iloc[:,1]+1)/2)
data['scaled_y'] = ((data.iloc[:,2]+1)/2)
data['scaled_z'] = ((data.iloc[:,3]+0.5))
data['x_bin']=pd.cut(x = data['scaled_x'],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
data['y_bin']=pd.cut(x = data['scaled_y'],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
data['z_bin']=pd.cut(x = data['scaled_z'],
bins = [p/100 for p in range(101)],
labels = [p for p in range(100)])
'''
#cmap = matplotlib.cm.get_cmap('Spectral') # Getting a list of color values
#data['color_dict'] = pd.Series({k:cmap(1) for k in data['scaled_x']})
#=====================PLOT DATA===========================#
# FOR BINS / GRID
if args.plot_type == "grid":
fig, ax = plt.subplots()
data.plot.scatter(x="x_bin",y="y_bin",c=0,colormap="Spectral", s=2, ax=ax)
plt.xlim(0,100)
plt.ylim(0,100)
elif args.plot_type == "3d":
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
xs = data.iloc[:,1]
ys = data.iloc[:,2]
zs = data.iloc[:,3]
ax.scatter(xs, ys, zs, marker="x")
else:
#fig, ax = plt.subplots(nrows=1, ncols=2)
fig, ax = plt.subplots()
# FOR JUST A SCATTER PLOT OF THE DESCRIPTORS - doesnt work for interactive selection
data.plot.scatter(x=1,y=2,c=0,colormap='viridis', s=2, ax=ax)
#data.plot.scatter(x=1,y=2,s=2, ax=ax[0])
#data.plot.scatter(x=3,y=4,c=0,colormap='viridis', s=2, ax=ax)
#data.plot.scatter(x=4,y=5,s=2, ax=ax[1])
plt.xlim(-0.5,0.5)
plt.ylim(-0.5,0.5)
# event to look out. visualization or closing the plot
fig.canvas.mpl_connect('key_press_event', lambda event: key_event(event, args))
fig.canvas.mpl_connect('button_press_event', lambda event: click_event(event, args))
plt.show()