Smart society for enjoying long healthy life

- Developing AI smart coaching technology

that facilitates voluntary skill-up for elderlies -

YUICHI KURITA

PROFESSOR, SCHOOL OF ENGINEERING,

HIROSHIMA UNIVERSITY

2

We would like to contribute to a smart society that

enjoys healthy longevity through AI technology that

supports training to make subjective skill up.

• In order to extend healthy life expectancy, training to

maintain physical functions and skills is necessary.

• To tackle sustainable skill-up training, it is important to

support independence and voluntariness.

• It is difficult to grasp personal skill.

• Training tasks are not based on evidence but are

subjectively determined by interpretation.

• The trainee can not realize the training effect.

Too difficult

Too easy Tired

Not motivated

Difficulty of voluntary skill-up support

When wrong tasks are given:

Collaboration with

• Retirement community for

elderly people (Sun City)

exists in the city.

• Needs for improving QOL

of elderly by facilitating

excise.

• 1st place of Most

Innovative Schools in U.S.

by News and World

Report (2015-17)

• Collaboration with Center

for Cognitive Ubiquitous

Computing (CUbiC)

6

The purpose of this research is to develop "AI smart

coaching technology to promote voluntary skill

enhancement" by simple and small number of

wearable sensors and individual skill grasping

technology.

AI smart coaching technology

1. Developing assessment system of user's

skill

2. Developing determination system of

task difficulty level

3. Developing smart coaching system

• Skill evaluation by wearable sensing, and motion

estimation by digital human model

• Proper task presentation by skill segment

• Development of artificial muscle training suits and training

software that can present task difficulty according to skills

Task

Dete

rmin

atio

n b

y AI

Effortestimation

PAM-basedtraining

Wearable sensing+

Digital human model

1. Developing assessment system of user's skill

Realize real-time skill segment

classification with a simple and

small number of sensors, and

visualizing the training effect.

Wearable sensing and skill

estimation by digital human modelWearable sensing

+Digital human model

Not-experienced Experienced

Movie provided by T. Terada lab, Kobe University

10Digital human model:

Dhaiba IMU (XSENS)

Movie provided by Digital Human Research Group, AIST

2. Developing determination system of task difficulty level

We will define physical effort from

physical load by using a

musculoskeletal simulation and set

the task level based on the physical

effort.

Level setting based on estimation of physical effort

Task

Dete

rmin

atio

n b

y AI

Effortestimation

3. Developing smart coaching system

Application to rehabilitation and special skill coaching and data accumulation

We will develop flexible and

lightweight artificial muscle suits,

training software with an intuitive

user interface, and verify the effect

for rehabilitation and special skills.

PAM-basedtraining

• Use lightweight, flexible, and inexpensive low-pressure

driven pneumatic artificial muscles (PAMs)

• A training suit that can be used at home as an extension

of clothes

• Maintain effective load by control the physical load

depending on the recovery condition of the user

Developing a PAM training muscle suit

加圧時 収縮

無加圧時

拘束部

空気受容部

人工筋：2本圧力：0.35MPaおもり：2kg

3. Developing smart coaching system

• A user interface that makes it possible to continue

rehabilitation and training that tends to be tired.

• Utilizing non-contact sensors realizes easy measurement

of the user motions

• Utilizing the developed AI core technology, AI will

automatically judge the tasks given to users depending

on individual skills

Developing a training software

3. Developing smart coaching system

Smart coaching technology with AI

1. Developing assessment system of user's skill

2. Developing determination system of task difficulty level

3. Developing smart coaching system

ＡＩ

データベース

1-1 Motion sensing by wearable sensors

1-2 Skill segmentationby using a DH

model

2. Task determination

3-1 Training software

3-2 PAM training suit

• In the lead research period, we will develop an effective

rehabilitation support system by linking the motor function data to

functional independence evaluation (FIM).

• The final goal is to develop an AI system that can be specifically

applied to special skills and to suggest work that will be made

possible by training.

Potential applications

1. Developing assessment system of user's

skill

2. Developing determination system of

task difficulty level

3. Developing smart coaching system

Toward smart society to enjoy healthy long life

We will contribute to a society that enjoys healthy longevity by:

• Improve independence of elderly and disabled people with technology that can

be used at home

• Extension of healthy life span by effective and efficient care for individual

• Reduce social security and medical expenses with securing manpower in aging

society

Task

Dete

rmin

atio

n b

y AI

Effortestimation

PAM-basedtraining

Wearable sensing+

Digital human model