nustep report at fefu
TRANSCRIPT
Sports, Simulation, AI and VRMonte Carlo Methods in Computer Simulations
of Complex Systems
NUSTEP Collaboration
FEFU, Vladivostok, Nov.11, 2016How to get Olympic Medals !
But neither Salmon Fishing nor Acorn Collecting Game
in Olympic . . .
National Geographics 2010© Itsu Horiguchi
Otomo Yasuo `Kuma-kun no
Aki’ Fukuin-Kan
2005
NUSTEP Collaboration
T.Asai, T.Hatsuda, R.Himeno, Y.Kawahara, S.Koike, H. Miyachi,
T.Miyosi, A.Molochkov, A.Nakamura, T.Ogi, T.Sakajo, T.Sumiya,
H.Takagi, M.Taki, N.Yamanaka
SportsSimulation Artificial Intelligence
Assimilation Virtual Reality
NUSTEP: Nextgeneration Unified Sport TEchnology Platform
2 /19
Leader
Plan of the Talk1. Blind Men touched Part of an Elephant
Sport Measurement Simulation with Assimilation AI (Artificial Intelligence) VR(Virtual Reality)
2. Target Sports Soccer Swimming Judo
3. International ? 4. Discussions
3 /19
1.1 Sports Measurement
Motion Capture
Wind TunnelAcceleration Meter
4
Water Tunnel
/19
1.2 Simulations
Men’sJumper� Women’sJumper�
Crosssec0on�Velocity,Surface:Pressure�
Latesepara0on
Earlysepara0on
FTFCE
FPE
(tendon)
(PE)
q
(CE)
αlT lMTC
lCE
FTFCE
FPE
(tendon)
(PE)
q
(CE)
αlT lMTC
lCE
Computational Fluid Dynamics
Modeling of Tendon-Frame5 /19
1.2 (Continue) Simulations+Assimilation
Assimilation? Cultural Assimilation?
All animals should eat Salmon !
Weather Forecast Mountain Fire
Rouchoux et al., 2014
6 /19
Data AssimilationSimulation + Real Measurement Data
Objective 1
Improvement of Simulation with the help of Real data.
Objective2
Compliment un-observed parts of data by physical
Simulation
Most Probable
Observed data
Different Initial/Boundary Conditions
K.Hosokawa, MTI-Handbook 7 /19
Tools for Data Assimilation
1. (Physical) Simulation Model
2. Measurement data
3. Statistical Science
4. High-performance Computers
8 /19
1.3 AI (Artificial Intelligence)
AI coach ?
What does a good coach do ?To know the difference between good and bad performance. (sometimes unconsciously)
Left, Right, Left,,,, Mmm.. Very Bad.
Great Progress mainly due to the Deep Learning.
9 /19
Let us find essential differences
between good and bad athletes
CC: j0sh (www.pixael.com)
CC: Agência Brasil
CC: R.H.Sumon
Input many sport measurement data (and data
assimilated simulations)
into AI. Tell me ‘good’ or ‘bad’ for each data.
10 /19
https://youtu.be/QZO9juIxptoAut.12, 2016
MIZUNO Corporation+Kawahara+Yamawaki
Expert Standard
11 /19
1.4 VR (Virtual Reality) for Educating PlayersMost Important, but most non-trivial parts.
We need more ideas.
We plan to compose a camera-image and VR.
12 /19
Play Oppositions in VR.
See one’s real movement and an ideal movement suggested by AI.
See an ideal movement from all direction views using a dataassimilated simulation.
or ..
How to use VR ?
13 /19
2.1 Soccer
Many studies in Tsukuba Univ. It includes
a Single Kick, CFD(Computational Fluid Dynamics)
Team Play.
14 /19
2.2 Swimming
Movement in the Water introduces new Research Points.
Many Studies in Tsukuba Univ.
15 /19
2.3 Judo
Depending Skill of Opponent. Good skill for one Opponent can
be Bad Skill for another.
In case of Soccer and Swimming, Skill is ‘closed’,
but for Judo it is ‘open’.
16 /19
3. International Collaboration
It is only for Japanese.
No, No. It is universal !
We want to collaborate with a university/institute out of Japan, to show NUSTEP is universal.
17 /19
Best partner ?There are Schools of ‘’Bio-Medicine’’, and “Arts, Culture and Sports’’ at FEFU. A member of NUSTEP, Dr. Yamanaka, is a Post-Doc at FEFU, who is an Asia Champion of Brazilian Jujutsu. with rank of Judo, 2-dan.
18 /19
4. Discussions
Now, this is only a Plan or our Dream. But all members believe this is worthwhile to pursue and this opens a new Research field. We appreciate any idea or proposal.
19 /19