technology innovation - hitachi
TRANSCRIPT
132
Development of Platform for Integrated Management of Value Chain for Regenerative Medicine Products
Because it involves multiple stakeholders along the
series of steps from extracting a sample from the patient
through to manufacturing, transportation, and admin-
istering the treatment back to the patient, regenerative
medicine demands stringent quality control and infor-
mation traceability. Accordingly, Hitachi has developed
a workfl ow system that utilizes IoT Compas to record
the progress of work as stakeholders hand on materials
and information from one to the other.
As samples taken from the patient are managed across
a number of branching processes, the resulting manage-
ment of workfl ow information has a complex structure
with multiple instances where steps branch or come back
together again. Nevertheless, regenerative medicine infor-
mation can be managed securely and with confi dence by
handling this data in IoT Compas with its ability to record
data interrelationships, and by providing fi ne-grained con-
trol of access to the information. Th e system itself it is able
to work with a variety of other systems by modifying the
data model that defi nes the workfl ow. Hitachi plans to
make it available to the pharmaceuticals industry as a data
management system for regenerative medicine.
1 Virtualization of Robot System Integration Using Collaborative Simulator
While there is rising demand for the automation of activ-
ities such as manufacturing and logistics, the diversity of
the work makes it more important than ever that system
integration (SI) coordinates and integrates the operation
of a wide variety of robots and other machinery to take
account of individual requirements. With this, there is
also a need not only to maximize the value delivered, but
also to make the design, approval, development, accep-
tance, and other steps faster, with allowance for people
being spread across diff erent remote locations without
access to the actual hardware.
It was with this in mind that Hitachi developed its
collaborative simulator, an integrated simulation environ-
ment. By providing a number of simulators intended for
diff erent purposes and applications, and by enabling tasks
such as the sharing and presentation of data, parameters,
and analysis results to be all tied together, analysis and
design is able to proceed in a comprehensive manner
that takes in various diff erent layers, including physi-
cal phenomena, the individual and group operation of
robots, and impacts on business. Th e simulator has also
been equipped with the ability to incorporate artifi cial
intelligence (AI) algorithms for things like the automatic
2
Technology Innovation:Industry
Order product Issue manufacturing and transportation instructions
Digital space (IoT Compas)
Patient information
Manufacturing and transportation
instructionsSample collection
recordsTransportation
recordsManufacturing
recordsAdministration
records
Take sample Transport sample Culturing/manufacturing
Transport product Administer
Medical institution
Pharmaceutical company
Medical institution
Transportation provider
Pharmaceutical company (overall coordination)
Manufacturer Transportation provider
Medical institution
1 Block diagram of workflow system using IoT Compas
IoT: Internet of Things
Hitachi Review Vol. 70, No. 3 350–351
Technology Innovation: IndustryResearch & Developm
ent
133
identifi cation of bottlenecks or for optimization achieved
through integration with IoT data.
Th e simulator has also been used under entirely remote
working conditions to design a system for demonstra-
tions at trade shows that combines three diff erent types
of robot, including development of the associated coor-
dination functions. Th e fi nal hardware integration work
conducted on-site was completed in a single day, with the
combined system successfully operating in accordance
with its design, without any need for rework. As a core
technology for merging the cyber and physical spaces, the
simulator will contribute to the deployment of advanced
automation systems.
Accuracy Improvement Service for Machining through Digitized Expert Know-how
Manufacturing that is independent of areas and people
is required in the trend of restructuring manufacturing
3
Three-robot systemSimulation in early stages of system integration planning Final simulation
Physical implementation
2 Example of planning, design, and implementation using collaborative simulator
The robot system at the center of the photograph was developed with support from a NEDO-funded project (Project for Innovative AI Chips and Next-Generation Computing Technology Development/Development of Innovative AI Edge Computing Technologies) (JPNP16007)NEDO: New Energy and Industrial Technology Development Organization
Manufacturing database
Operator
Machine tool
Original NC data
Modified NC data
MaterialsCutting tools
Machine tools
Manufacturing experience database
Request digital recipe
generation
Digital recipe generation for machiningCutting tool
Workpiece
Output digital recipe
Tool position (actual)
Example of modified NC dataTool position (design value)
Machine stiffness
Tool deflection = Cutting force/Tool stiffness
Tool stiffness
Machining direction Tool deflection
Cutting forceCutting force
Workpiece
Assess & analyze
User
Machine individuality (stiffness) data
Manufacturing experience
Accuracy improvement service for machining (web-based cloud service)
Module for digital recipe generation for machining
Machining simulation that takes account of machine individuality (stiffness)
Use
Modify NC data based on tool deflection
Compensate tool deflection (generate modified NC data)
Correction
3 Overview of accuracy improvement service for machining
134
bases accelerated by COVID-19 and international rela-
tions between major powers. To realize such a situation,
product quality has to be stabilized regardless of the pro-
duction equipment and skilled workers there. Th erefore,
Hitachi has developed a technique for digital recipe-
generating technologies that digitize expert know-how
and optimize process conditions input into production
equipment. Hitachi provides an accuracy improvement
service for machining, which is one of the major process-
ing methods, as a web-based cloud service.
In this technology, cutting tool defl ection is accurately
calculated by machining simulation incorporating the
individuality of each machine tool, and numerical con-
trol (NC) data is modifi ed to compensate the calculated
machining errors. By digitizing these machine individu-
alities, which skilled workers hold as their own know-how,
the service realizes accurate machining regardless of the
machine tools and skilled workers. In the future, Hitachi
plans to add services for digital recipe-generating tech-
nologies for other processes such as plastic forming and
injection molding.
New Oil-fl ooded Screw Air Compressors with Industry-leading Energy Effi ciency
Hitachi has commenced mass-production of two new
oil-fl ooded screw air compressors (22 and 37 kW) that
adopt its proprietary impinging oil atomization mecha-
nism. Th e new compressors are the fi rst in the world*
to feature this mechanism whereby impingements
between lubricating oil causes it to atomize, thereby
promoting cooling of the compressed air by the lubri-
cant and reducing compression power. Th e result is
* Th is solution is supplied by Hitachi Solutions, Ltd.
4
world-leading energy effi ciency [6.4 kW/(m3/min), an
approximate 6% improvement on the previous model],
that satisfi es the GB1 class in China’s energy effi ciency
regulations.
Along with an approximate 1.5-fold increase in rated
operating speed, the compressor has also been downsized
by about 10% by reducing the displacement and switch-
ing to a high-speed/low-torque motor. In addition to the
product’s energy effi ciency, these measures also encourage
optimal operation of the entire compressed air supply
system by allowing compressors to be spread across the
site, contributing to a reduction in plant carbon dioxide
emissions.
Future plans include supplying the compressors to
markets outside Japan, taking advantage of the standard-
ization of core components with Sullair, LLC.
(Hitachi Industrial Equipment Systems Co., Ltd.)
Discussion Paper on Revitalizing Human-machine Interaction in Digital Society Published in Collaboration with acatech
Given the greater workforce diversity that comes with a
falling population and the greater diversity in machines
due to technologies like the IoT and AI, industry too
is seeing increasing variety in its humans and machines.
For this to bring sustainable economic growth calls
for a world where humans and machines can evolve
in tandem.
acatech, the German Academy of Science and
Engineering, has established a project that addresses this
challenge and is led by Professor Henning Kagermann,
one of the people behind Germany’s Industrie 4.0 vision,
with Yoichi Nonaka, a Senior Chief Researcher at
Hitachi’s Research & Development Group, as co-
leader. Th e results of debate by experts from Japanese
and German industry and academia were published
in September 2019 in a discussion paper entitled,
“Revitalizing Human-Machine Interaction for the
Advancement of Society.”
Th is paper described the need for a digital knowledge
platform shared by humans and machines so as to pro-
vide ways in which they can uplift one another. Giving
the name “Multiverse Mediation” to this platform, this
paper also presented examples from Japan and Germany
of fl exible industrial systems that are able to achieve the
best overall outcome by having a wide variety of humans
* As of October 2020, based on research by Hitachi Industrial
Equipment Systems Co., Ltd.
5
4 New oil-flooded screw air compressors (37 kW)
Hitachi Review Vol. 70, No. 3 352–353
Technology Innovation: IndustryResearch & Developm
ent
135
and machines make up for each other’s strengths and
weaknesses.
Th e project is currently expanding the scope of debate
and engaging in activities to help create a society in which
humans and machines coexist.
5G-powered Real-time Edge AI Framework for Realizing Effective Remote Robot Control
Today’s trend of multi-tiered systems requires IoT system
to integrate with multiple technologies and co-exist with
6
Former Today Future
Multiverse MediationTradition
Support
Machine Machine Machine
Operation
Human
Human Human Human
lti
nen MaMutual Aid
ana
Mutual Aid
Mul MedM l
Mutual Aidersetion Mutual Aid
Human
5 Evolution of human-machine interaction
5G-powered interactive remote collaboration system between skilled workers and robots
Manufacturing Railway
Real-time edge analytics (ML/AI) Robotics platform
ML/AI models/field devices management
5G infrastructure
Energy
Data fusion/data synchronization/data curation
TSN
Smart city/space
Flexible automation/optimization
Mobility maintenance/management Smart grid COVID-19
Verticals
AnalysisIoT platform
5G
Control edge
Use cases
Data
Connectivity
Physical field
Object detection/recognition
Unstructured Structured
Anomaly detection
Acoustic Vibration
Robots Human Machines Environment
Video LiDAR Motion PLC MES ERP
Machine failure detection
5G
Real-time edge AI in MEC
Large-capacity sensing data
4K/TOF camera
Robot arm5G setups
Control signal
4K video
Instruction
Remote worker
Management agent
Interactive UI
Trajectory optimization
Grasp planning/ranking
Management agent
Human/human action recognition
Signal denoising
6 Framework for 5G-powered real-time edge AI solutions, and its application to remote robot control
MEC: multi-access edge computing ML: machine learning PLC: programmable logic controller TSN: time sensitive network MES: manufacturing execution system ERP: enterprise resource planning TOF: time of fright UI: user interface LiDAR: light detection and ranging * Function blocks surrounded by the red rectangles indicate utilized in the application.
136
with control cycle times in the order of hundreds of mil-
liseconds. Th is makes it impractical for use with control
cycles of just a few milliseconds, which is what is needed
to operate robots safely given the external factors found
in the real world, such as fl oor vibration or arm collisions.
In response, Hitachi has developed a high-speed
controller system made up of a small edge controller for
executing control acquired from deep learning; a small
input/output (I/O) controller to handle sensor, motor, and
other I/O; and a real-time network linking these together.
High-speed processing is achieved by implementing it
in hardware circuits on fi eld-programmable gate arrays
(FPGAs) in the controllers. Testing has demonstrated
that this can achieve a 5-ms response time for reacting to
external events. Th e intention is to utilize the system in
actual applications equipped with these controllers.
System for Formulating and Managing Optimal Product Space Allocations and Process Plans in Factories with Wide Product Range
Because short-run manufacturing requires that the same
equipment be used for diff erent products, with those
products having diff erent sizes and taking up diff ering
amounts of space in the production process, there is
a need to formulate process plans that allow for effi -
cient space allocation and equipment use sequencing.
Unfortunately, the complexity and myriad permutations
of space and equipment use has in the past meant that
coming up with these plans has required a lot of work
by expert staff .
8
other solutions as well as partner/vendor ecosystem. To
enable this scalable solution architecture for multiple
verticals, Hitachi America, Ltd. has established a soft-
coupling approach in the fi fth-generation (5G)-powered
framework for real-time edge AI and robotics.
In order to perform realistic validation of this frame-
work, the company established a real-time and interactive
remote collaboration system between skilled workers and
robots in close collaboration with Georgia Institute of
Technology. Th e core function of this system is real-time
edge AI extracting insights not only from large-volume
sensing data relayed over dedicated 5G network, but also
human interaction data collected via intuitive human
machine interface (HMI) which, with support of 5G’s
enhanced mobile broadband (eMBB) feature, drastically
improves response time and usability for remote workers.
Th is validates the capability of the framework to explore
new age of fl exible automation in manufacturing industry,
as well as possibility of this framework to scale multiple
verticals.
(Hitachi America, Ltd.)
High-speed Controller System with Hardware-based Network Control and Deep Learning
Amid growing demand for robots as a solution to labor
shortages, progress is being made on ways of controlling
them without the need for programming by using deep
learning to provide robot operation control. Th e prob-
lem with simply combining deep learning control with
existing controllers, however, is that this is only realistic
7
60 mm
30 mm
FPGA
FPGA
FPGA
FPGA
FPGA
ExampleDetection of bad connections in wiring harnesses
Information from sensorssors
High-speed feedback
Externalfactor
Externalfactor
Realtime networkSmall edge controller
100 µs
Input data
Input data
Small I/O controller
Install close to equipment
Sensor
Sensor
Sensor
Sensor
Reflexive response(avoidance, repositioning, etc.)
Networkcontrol
Deep learningexecution
Model(pre-taught)
7 Block diagram of new control system
Hitachi Review Vol. 70, No. 3 354–355
Technology Innovation: IndustryResearch & Developm
ent
137
new control technology takes advantage of the fact that
the wire rope length is at its shortest when there is no
such misalignment. It works by winding the wire rope
before lifting and measuring its length at the point where
all of the slack is taken in and the wire rope becomes
tensioned, moving the trolley to reduce the misalign-
ment. To detect when the wire rope becomes tensioned,
Hitachi also developed a load estimation method using
deep learning to estimate the weight of the suspended
load on the trolley. Th is is able to estimate the weight to
within 50 kg and enables tensioning of the wire rope to be
detected with high accuracy. By using these technologies,
load sway at liftoff that would have been around 200 mm
with manual operation was reduced to less than 100 mm.
In the future, Hitachi intends to further improve safety
by developing even more precise control technology.
In response, Hitachi has devised an algorithm that
eliminates delays due to lack of space by planning where
to put products while also taking account of the schedule
of equipment use. A scheduling technique that auto-
matically generates optimal process plans has also been
developed. It can be used to investigate ways of keeping
plans on track, being able to show the progress of work
in the factory by cross-checking image data from cameras
installed in the plant against the plan.
An evaluation of the new system made using historic
operational data from a short-run manufacturing plant
indicated that it could reduce work time by 20%.
Control Technology for Adjusting Initial Position of Hoist Crane
Th e labor shortage caused by the aging of skilled workers
and rising demand for cranes is increasing the number
of unskilled workers. Because such workers are unfamil-
iar with how to prevent load sway, this brings a height-
ened risk of collision, catch, and other such accidents,
and causes an increase in work time. Accordingly, with
the aim of improving safety and shortening work times,
Hitachi has developed a control technology for adjusting
the initial position of the hoist cranes that suppresses the
load sway that occurs when a suspended load is lifted off
the ground.
Th is problem of load sway at liftoff occurs if there
is any horizontal misalignment between the top of the
suspended load and the trolley used to transport it. Th e
9
Work schedule (Gantt chart)
Lot: DK30759Customer: XXXXXXProduct: XXXXXXInward goods: 4/19Dispatch: 6/26Start work: 4/19Completion: 5/17Duration: 28 days
Lot: DK30741-4Customer: XXXXXXProduct: XXXXXX
Lot: DK30741-4Customer: XXXXXXProduct: XXXXXXPart: LIM-2Process: Assembly
STEnlargement
Enlargement
B1
A13
Lot: DK30741-4Part: LIM-2Process: Preliminary drying
A6
A15A11
B2
B1
B1B2
B1
A7
A15
A13
A11A7
A6
Span
use
d
4/24 4/24Clean-
ing
MeasurementCoating
Cleaning
Drying
CoatingTidy upTidy up
Drying
Salt removalRT
BGBKLBGBKR
CLRETC1ETC2IM2
Space dataProduct data
Output
Presentation
Worker Simultaneous optimization
Image dataIn-plant camera Image comparison and analysis
Feedback
Data entry
Designer
Layout schedule
Layout Camera images
(1) Simultaneous optimization of work schedule and product layout
(2) Output of process planning
(3) Work progress presentation
8 Overview of system for formulation and management of process plans
Move trolley
No
Yes
Yes
No
Winding wire rope
Is wire rope tensioned?
(determine by deep learning)
Measure wire rope length
Load
Wire rope
Trolley
Initial position adjusting complete
Is wire rope length at minimum?
9 Control technology for adjusting initial position of hoist crane