Pre-Post Analysis on Multi-Skill Development using Flow Line Data at Expressway Service Area Facilities

This paper mainly reports on a case study of pre-post analysis of multi-skill development at commercial facilities in an expressway service area, together with the results of grasping the current work patterns of the site. A work pat-tern analysis method that does not depend on site-specific detailed information using geospatial intelligence (GSI) technology was applied to the facilities, and a matching analysis with site-specific detailed information was also conducted. In the facilities, 16 work patterns were extracted that are close to the actual situation. In addition, it was confirmed that the number of work areas that can be handled has increased as a result of multi-skill development. The median number of employees per day was reduced by one after multi-skill development, and its standard deviation was also reduced. This implication is that the site became more flexible to handle daily tasks. In contrast, walking distances and work hours per employee increased. Further improvement in efficiency and automation will be desirable to absorb the reduction impact in the number of employees without increasing the burden on each individual. In this case study, the analysis method developed for a manufacturing line was applied to the service site. This provided a preliminary validation of the versatility of this analysis method and its applicability to larger sites.

1. Pre-Post Analysis on Multi-Skill Development
using Flow Line Data
at Expressway Service Area Facilities
Takeshi Kurata
1
,
Akihiro Sato
1
, SatokiOgiso
1
, Karimu Kato
1
, Satoshi Nakae
1
,
Ryosuke Ichikari
1
, Takeshi Shimmura
2
1
Human Augmentation Research Center, AIST, Japan
2
GANKO Co. Ltd., Osaka, Japan
APMS 2024

2. Objectives of this case study
2
•Objective and quantitative comprehension of work
patterns of employees from flow line data at
commercial facilities.
•Confirmation of the effectiveness of multi-skill
development.
•Preliminary verification of the versatility of the work
pattern analysis method developed for automotive
parts manufacturing lines.
•Out of scope of this presentation:
Prediction of
physical and mental status for always-on monitoring in
health care services, including human capital management support.

3. Work site of this case study
3
Site: Commercial facilities in Suita Service Area (SA) of the Meishin Expressway,
Japan
Measurement period:
First period: Jul 4 - Jul 14, 2022
Second period: Jan 16 - Jan 26, 2023
Total 3,600 hours
No data in 10% due to battery issues
Notes:Multi-skill development was implemented between the first and second
measurements for each employee to be able to carry multiple positions.
Area of measurement: Approx. 2,200m
2
Participants: Employees
First measurement: 48
Second measurement: 42
(37 in both measurements)

4. Extracted exceptions of work area transition instances
4
Seven extracted exceptions of work area
transition instances along with their
exception indicators and the threshold.
Exceptions A and B for visual inspection.
5
Kurata, T., Watanabe, R., Ogiso, S., Mori, I., Miura, T., Kato, K., Haga , Y., Hatakeya-ma, S., Kimura, A., Nakahira, K., Work Pattern Analysis with and
without Site-Specific Information in a Manufacturing Line, APMS, IFIP AICT 689, pp. 253- 266 (2023)
@APMS 2023

5. Outline of work pattern analysis method
without site-specific information
5
Kurata, T., Watanabe, R., Ogiso, S., Mori, I., Miura, T., Kato, K., Haga , Y., Hatakeya-ma, S., Kimura, A., Nakahira, K., Work Pattern Analysis with and
without Site-Specific Information in a Manufacturing Line, APMS, IFIP AICT 689, pp. 253- 266 (2023)
@APMS 2023

6. This presentation: Analysis
from flow line data
Integrated positioning system:
•BLE RSSI
•PDR
•Floor plan data
Measurement for multi-objective case study
6
In blue frames: positioning-related items
T. Kurata et al., “Project Progress on XR-AI Platform for Tele-Rehab and Health Guidance,” 2024 IEEE VR Abstracts and Workshops (VRW), pp.244-
251, 2024.
RSSI (received signal strength)
PDR (pedestrian dead reckoning)
Why flow line?

7. Geospatial Intelligence (GSI) with 6M data
7
IE: Industrial Engineering, OR: Operations Research,
IoT: Internet of Things, IoH : Internet of Humans, UI: User Interface
XR: VR, AR, MR, etc. (AR: Augmented Reality, VR: Virtual Reality, MR: Mixed Reality)
RM: Raw Material, WIP: Work In Progress, SFG: Semi-Finished Goods, FG: Finished Goods
Conceptual diagram of GSI 6M information in Service and
Manufacturing sites

8. HPM/HCM and Geospatial Intelligence (GSI)
8
Kurata, T., Geospatial Intelligence for Health and Productivity Management in Japanese Restaurants and
Other Industries, APMS, pp. 206–214 (2021) doi: 10.1007/978-3-030-85906-0_23

9. Floor plans of Suita SA commercial facilities
9

10. Work area transition model from flow line data
10
Histogram of stay length (length of elapsed time per
stay at a stay node) and the clustering result.
29 stay nodes:
28: in the facility
1: group of all other nodes
9 levels of stay length at stay nodes
Result

11. Representative instances of work patterns
11
← 16 work patterns:
Frequency of occurrence (person-day) by
role (site-specific detailed information).
extracted by
clustering 415 work instances
"work area transition instance," or
"work instance" for short:
obtained by assigning move/stay
probabilities regarding one continuous
flow line data for one employee
Result

12. Pre-post comparison of the distribution of 16 work patterns
12
3D plots using the first to third principal component axes.
•The number in each plot: the work pattern ID
•The size of the plot: the amount of work instances belonging to the corresponding work pattern.
Result

13. Pre-post comparison of representative instances of major
work patterns
13
Work area transition model
Excluded from
this analysis
Result

14. 14
Pre(2022.7) Post (2023.1)
Multi-skilldevelopment
(from 2022.11)
Group A: First Floor
•All or part on the first floor consisting of the
food court seating area, kitchen, shopping
area, cash register, and hot snack stand
Wide area of the first floor
Result

15. 15
Pre(2022.7) Post (2023.1)
Multi-skilldevelopment
(from 2022.11)
Group B: Restaurant plus first floor
•Restaurant on the second floor (seating,
kitchen, entrance)
•Kitchen of the food court on the first floor
•Restaurant on the second floor
(kitchen and entrance reduced)
•Wide area of the first floor
Result

16. Group C:Entire first floor plus Restaurant kitchen
16
•Entire first floor
•Restaurant kitchen on the second floor
Result
Post (2023.1)

17. Summary of pre-post multi-skill development
17
Pre PostDiff.Diff. [%] Stat. diff.
Avg. # of work areas/day (all)11.113.3 2.2 19.8 Up, no test
(participation in both)11.313.21.9 16.8 Up, p≒0.01
Walking distance [km/person-day] 4.6 5.00.5 10.2 Up, p≒0.036
# of non-routine instance 9.021.012.0 133.3 Up, no test
# of employees/day [median]24.322.1-2.2 -9.1 Down, p≒0.036
[SD] 2.8 1.5-1.3 -46.3 Down, no test
Avg. hourly wage (July as 1) 1.001.030.03 3.00 Even, no test
Avg. sales/day (July as 1) 1.001.020.02 2.30 n.s., p≒0.747
Avg. length of work hours/person-day (July as 1)1.001.070.07 6.60 Up, no test

18. Comparison of two case studies
18
Auto parts mfg. line Expressway SA Facilities
# of all stay nodes 21 29
# of stay nodes for analysis 13 28
# of stay length levels 4 9
Model dimensions 525 1,102
46 415
12 16
7 30
15.2 7.2
Meas. area [m
2
] 1,400 2,200
# of BLE beacons 56 97
Days of meas. 5 20
# of subjects 10 53
Etc.
# of roles 6 24
# of work areas for analysis 10 25
Etc.
Floor plans, BLE beacon layout
site-specific
detailed
information
Work area classification, Shift , Attendance records,
Outputs (# of productions, sales)
Work area
transition
model
# of work instarnces
# of work patterns
# of non-routine instances
non-routine ratio [％]
Information for
measurement/
prepapration
non-routine ratio: the ratio of the number of non- routine work instances to the total number of work instances
•manufacturing site to service site
•Larger site
APMS 2023 APMS 2024

19. Conclusions
19
Objectives:
•Objective and quantitative comprehension of work patterns on the
site.
•Confirmation of the effectiveness of multi-skill development
implemented between the first and second measurements.
•Preliminary verification of the versatility of the work pattern analysis
method developed for automotive parts manufacturing lines.

20. Conclusions
20
Objectives:
•Objective and quantitative comprehension of work patterns on the
site.
•Confirmation of the effectiveness of multi-skill development
implemented between the first and second measurements.
•Preliminary verification of the versatility of the work pattern analysis
method developed for automotive parts manufacturing lines.
•The managers and executives understood how the multi-skill development was
planned.
•Difficult to obtain quantitative and objective results on how each
employee's work style had changed and the extent of the change.
•This analysis: Uncovered the above.
•IT and data generally flatten the organization
•Further co-creative improvements will be investigated on this basis.

21. Future Works
21
The positioning error (2.5m): Within the acceptable range
•Size of each work area is relatively large compared to the positioning error.
•Analysis results clearly confirm the change in work patterns due to multi-skill
development.
•Difficult to perform work analysis at a finer granularity, such as at which table the
work was executed.  Other positioning methods such as UWB (Ultra-Wide
Band) or computer vision should be adopted.
Mapping between work areas and work tasks
•When different tasks are performed in the same area, such as in office work,
another mechanism for mapping measurement data and work tasks is necessary.
More comprehensive analysis
•Combining biological data (pulse-rate-related data), and subjective data
(questionnaires, interviews, and experience sampling using emoji) with location
information.
Work pattern simulators
•Introducing the work area transition model and work patterns in simulation for
pre-evaluation of site improvement/design ideas.

22. 22

23. Floor plans of Suita SA commercial facilities
23

24. Work pattern analysis w/ GSI:
Start with or without site-specific information?
24
Site-specific detailed information: Non- measured information obtained through interviews and inquiries at the site, such
as each employee's role, detailed shift plans, and typical work area classifications

25. Pre-post comparison of representative instances of major
working patterns
25
Work area transition model
Excluded from
this analysis
Result
•All or part on the first
floor consisting of the
food court seating area, kitchen, shopping area,
cash register, and hot
snack stand
Wide area of the first floor
•Restaurant on the second floor
(seating,
kitchen, entrance)
•Kitchen of the food
court on the first floor
•Restaurant on the
second floor(kitchen
and entrance reduced)
•Wide area of the first floor
•Entire first floor
•Restaurant kitchen on the
second floor