Slide 1
Slide 1 text
4 D 29 26
1 . 0 I
1
Slide 2
Slide 2 text
& .-* (2
,0'/4"# 51
83;7 +)
!& % $ 9(
6:
Attention
Slide 3
Slide 3 text
3
#cmdevio2019
Slide 4
Slide 4 text
4
os
t m (
L @S
g E b
i _d L rMI D
@ E
( (
( ) (
e
a
n
k AWS
E
Slide 5
Slide 5 text
5
DeepRacer
Slide 6
Slide 6 text
6
D
Slide 7
Slide 7 text
7
)
(
Slide 8
Slide 8 text
8
…
Slide 9
Slide 9 text
9
DeepRacer
4 D 26
9 01 .
Slide 10
Slide 10 text
10
DeepRacer
A A
A
Slide 11
Slide 11 text
11
1
2
3
Slide 12
Slide 12 text
12
DeepRacer
Slide 13
Slide 13 text
13
Slide 14
Slide 14 text
14
DeepRacer
1/18
3D
AWS DeepRacer League
Slide 15
Slide 15 text
15
DeepRacer
https://aws.amazon.com/jp/deepracer/
Slide 16
Slide 16 text
16
DeepRacer
! &%$
+)*2
1
'/*2
(#-,
0.
"
Slide 17
Slide 17 text
17
3D
AWS RoboMaker
Robot Operating System (ROS)
Gazebo rqt
Slide 18
Slide 18 text
18
AWS DeepRacer League
⁻ 0 1 : 9 A
⁻ 9 2 R
⁻
⁻ D I
⁻ 1 2
⁻
https://aws.amazon.com/jp/deepracer/league/
Slide 19
Slide 19 text
19
Slide 20
Slide 20 text
20
(Artificial Intelligence, AI)
(Machine Learning, ML)
NeuralNetwork
DeepLearning
Slide 21
Slide 21 text
21
Slide 22
Slide 22 text
22
=
1
(
) ( (
Slide 23
Slide 23 text
23
L
N
-
) ( - D
Q
Slide 24
Slide 24 text
24
DeepRacer
Cliped PPO
PPO (Proximal Policy Optimization)
OpenAI2017
Slide 25
Slide 25 text
25
( (
) )
Slide 26
Slide 26 text
26
1
Slide 27
Slide 27 text
27
) ()
(
Slide 28
Slide 28 text
28
DeepRacer
Slide 29
Slide 29 text
29
DeepRacer
+
+ +
Slide 30
Slide 30 text
30
DeepRacer
Slide 31
Slide 31 text
31
DeepRacer
…
Slide 32
Slide 32 text
32
DeepRacer
Slide 33
Slide 33 text
33
orz
Slide 34
Slide 34 text
34
DeepRacer
+
+ +
Slide 35
Slide 35 text
35
DeepRacer
D
A
D
Slide 36
Slide 36 text
36
$ '
+
( #
&!
%"
Slide 37
Slide 37 text
37
(
) )
https://docs.aws.amazon.com/ja_jp
/deepracer/latest/developerguide/
deepracer-train-models-define-
reward-function.html
Slide 38
Slide 38 text
38
Slide 39
Slide 39 text
39
⁻ 10
⁻ :
Slide 40
Slide 40 text
40
SageMaker RL + RoboMaker
Slide 41
Slide 41 text
41
SageMeker RLRoboMakerGA
Slide 42
Slide 42 text
42
SageMaker “RL”
⁻
⁻ M
⁻ M M
⁻ M
⁻
⁻
⁻ J M S
Slide 43
Slide 43 text
43
DeepRacer )
D
( )
)
(
)
Slide 44
Slide 44 text
44
SageMaker
https://dev.classmethod.jp/machine
-learning/sagemaker-robomaker-
deepracer-sample/
Slide 45
Slide 45 text
45
$#
"
!
https://github.com/awslabs/amazon-sagemaker-examples
Slide 46
Slide 46 text
46
Jupyter
!
Slide 47
Slide 47 text
47
( (
)
Slide 48
Slide 48 text
48
Slide 49
Slide 49 text
49
2
1
Slide 50
Slide 50 text
50
2
2
Slide 51
Slide 51 text
51
(.
( )(
Slide 52
Slide 52 text
52
$2 #
"
/1(+ $2 #,- !
https://docs.aws.amazon.com/ja_jp/deepracer/latest/developerguide/deepracer
-iteratively-enhance-reward-functions.html
*
$2%)
'.0&
)
Slide 53
Slide 53 text
53
Best Practices when training with PPO (Unity Technologies)
https://github.com/Unity-Technologies/ml-agents/blob/master/docs/Training-PPO.md
Slide 54
Slide 54 text
54
DeepRacer "%
!
"%
$#
!
!
Slide 55
Slide 55 text
55
DeepRacer
Slide 56
Slide 56 text
56
DeepRacer
Slide 57
Slide 57 text
57
DeepRacer
Slide 58
Slide 58 text
58
DeepRacer
Slide 59
Slide 59 text
59
DeepRacer
Slide 60
Slide 60 text
60
Slide 61
Slide 61 text
61
• g
• + +
• M D c
• R S D a k
• b
• LL e
Slide 62
Slide 62 text
62
DeepRacer
Slide 63
Slide 63 text
No content