forked from Bharath2/iLQR
-
Notifications
You must be signed in to change notification settings - Fork 0
/
containers.py
130 lines (105 loc) · 3.52 KB
/
containers.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
import sympy as sp
import numpy as np
from numba import njit
from .utils import *
class Dynamics:
def __init__(self, f, f_x, f_u):
'''
Dynamics container.
f: Function approximating the dynamics.
f_x: Partial derivative of 'f' with respect to state
f_u: Partial derivative of 'f' with respect to action
f_prime: returns f_x and f_u at once
'''
self.f = f
self.f_x = f_x
self.f_u = f_u
self.f_prime = njit(lambda x, u: (f_x(x,u), f_u(x,u)))
@staticmethod
def Discrete(f, x_eps = 1e-4, u_eps = 1e-4):
'''
Construct from a discrete time dynamics function
'''
f = njit(f, cache = True)
f_x = njit(lambda x, u: FiniteDiff(f, x, u, 0, x_eps))
f_u = njit(lambda x, u: FiniteDiff(f, x, u, 1, u_eps))
return Dynamics(f, f_x, f_u)
@staticmethod
def SymDiscrete(f, x, u):
'''
Construct from Symbolic discrete time dynamics
'''
f_x = f.jacobian(x)
f_u = f.jacobian(u)
f = sympy_to_numba(f, [x, u])
f_x = sympy_to_numba(f_x, [x, u])
f_u = sympy_to_numba(f_u, [x, u])
return Dynamics(f, f_x, f_u)
@staticmethod
def Continuous(f, dt = 0.1, x_eps = 1e-4, u_eps = 1e-4):
'''
Construct from a continuous time dynamics function
'''
f = njit(f)
f_d = lambda x, u: x + f(x, u)*dt
return Dynamics.Discrete(f_d, x_eps, u_eps)
@staticmethod
def SymContinuous(f, x, u, dt = 0.1):
'''
Construct from Symbolic continuous time dynamics
'''
return Dynamics.SymDiscrete(x + f*dt, x, u)
class Cost:
def __init__(self, L, L_x, L_u, L_xx, L_ux, L_uu, Lf, Lf_x, Lf_xx):
'''
Container for Cost.
L: Running cost
Lf: Terminal cost
'''
#Running cost and it's partial derivatives
self.L = L
self.L_x = L_x
self.L_u = L_u
self.L_xx = L_xx
self.L_ux = L_ux
self.L_uu = L_uu
self.L_prime = njit(lambda x, u: (L_x(x, u), L_u(x, u), L_xx(x, u), L_ux(x, u), L_uu(x, u)))
#Terminal cost and it's partial derivatives
self.Lf = Lf
self.Lf_x = Lf_x
self.Lf_xx = Lf_xx
self.Lf_prime = njit(lambda x: (Lf_x(x), Lf_xx(x)))
@staticmethod
def Symbolic(L, Lf, x, u):
'''
Construct Cost from Symbolic functions
'''
#convert costs to sympy matrices
L_M = sp.Matrix([L])
Lf_M = sp.Matrix([Lf])
#Partial derivatives of running cost
L_x = L_M.jacobian(x)
L_u = L_M.jacobian(u)
L_xx = L_x.jacobian(x)
L_ux = L_u.jacobian(x)
L_uu = L_u.jacobian(u)
#Partial derivatives of terminal cost
Lf_x = Lf_M.jacobian(x)
Lf_xx = Lf_x.jacobian(x)
#Convert all sympy objects to numba JIT functions
funs = [L, L_x, L_u, L_xx, L_ux, L_uu, Lf, Lf_x, Lf_xx]
for i in range(9):
args = [x, u] if i < 6 else [x]
redu = 0 if i in [3, 4, 5, 8] else 1
funs[i] = sympy_to_numba(funs[i], args, redu)
return Cost(*funs)
@staticmethod
def QR(Q, R, QT, x_goal, add_on = 0):
'''
Construct Quadratic cost
'''
x, u = GetSyms(Q.shape[0], R.shape[0])
er = x - sp.Matrix(x_goal)
L = er.T@Q@er + u.T@R@u
Lf = er.T@QT@er
return Cost.Symbolic(L[0] + add_on, Lf[0], x, u)