PDF version of this report
You must have Adobe Acrobat reader to view, save, or print PDF files. The
reader is available for free
download.
Swarm Robotics for Manufacturing
Copyright 2020, Faulkner Information Services. All
Rights Reserved.
Docid: 00021378
Publication Date: 2003
Report Type: TUTORIAL
Preview
The term ‘swarm robotics’ is indicative of a revolution in the field of
manufacturing. This technology offers a largely autonomous flock of robots that
are usually sized no bigger than a shoebox. The area is frequently driven by
advancements in swarm behavior control software and wireless mesh networks, and,
in turn, can embed and control many individual robots to orchestrate coordinated
action. Generally, swarming robotics is regarded as a game changer for manufacturing at
every scale; it also could ultimately speed the demise of ‘big’ factory-floor
human workforces. This tutorial examines the ‘swarm robotics’ phenomenon.
Report Contents:
Executive Summary
[return to top of this
report]
The field of robotics is rapidly entering into a new dimension: Swarm
robotics. This revolution features largely autonomous flocks of robots –
typically sized between that of a walnut and a
shoebox – and is driven by advancements in swarm
behavior control software and wireless mesh networks. This technology, in
turn, can embed and
control many individual robots to orchestrate coordinated action.
Most robots built for swarming are adroit but simple. Conversely, the
wireless mesh network that connects both controller to robot and
individual robots to one another provides individualized, sophisticated,
task-oriented intelligence. Locating intelligence outside robots means an
entire robotic system can be readily moved, updated, adapted, or
repurposed. A robot or group of robots can have a different jobsite and
even a different job from one deployment to the next.
The robot piece of a swarm solution is typically small, durable, and highly
mobile. Robots come in various configurations and feature motility strategies
such as:
- Airborne – Maintain synchronized flight, dynamically change
direction without latency, interacting as if a flock of birds. - Crawler – Move rapidly over uneven surfaces, even scaling
vertical barriers and flowing over or around obstacles. - Ball Bots – Can roll, bounce, and fly.
Wireless mesh networks are also a key element of swarm robotics. In
real-world applications this type of connectivity offers a robust solution for
areas such as
manufacturing, surveillance, and process control. These options can:
- Scale, are fault tolerant, and are flexible.
- Can operate in the global frequency band
reserved for license-free use in industry, science, and medical
applications. - Are deployable worldwide.
- Can embed security into robotic control meshes.
- Employ multiple compartmentalized, durable, and redundant
security strategies.
Swarms can also be intelligent enough to carry out their functions
without
external supervision once tasked. These intelligent swarms can operate
indoors or outdoors, orienting themselves and moving through space without
reliance on GPS navigation. They are extremely fast, maneuverable, and
agile, and can work in teams. In addition, they can rapidly map, measure,
and evaluate environments or even lift, move, and assemble objects. It is
not unreasonable to expect to see the commercial application of swarming
robotics available as turn-key systems in the next two to five years.
Description
[return to top of this
report]
Robots have been used in heavy manufacturing for more than fifty years.
Benefits
Robot swarms are free to move, are built at a human-friendly scale, are
highly flexible in terms of tasking, and can dynamically adapt to the presence
of people and objects. Manufacturing is on the threshold of a revolution built on a new
generation of robotic workers. In contrast to older types of master / slave robotic control
systems, wireless mesh-based swarm robotics are designed to move
freely and accomplish tasks with little or no human
intervention after initial tasking. A typical manufacturing swarm robotic
solution consists of three basic layers:
-
Software applications that provide intelligence and task scripting
for the work the robot swarm will perform. -
Wireless mesh networks that embed and control many robots,
orchestrating their actions and providing communication between
individual robots and with backend
supervisor software. -
Robots that are small, light, mobile, and
available, off the shelf.
Wireless Communication + Mesh Topology = Networking Magic. The state of the art in robotics today is driven by rapid
refinement of wireless networks based on the
concept of mesh network topology. Mesh
topology allows the location of network elements to move and the
membership of the network to dynamically change, grow, or shrink.
A mesh
network’s defining characteristics include:
- Decentralized and flat, in terms of
command and control hierarchy. - Aware mesh nodes that can collect and share
information between them. - Self-organizing and, in the event of individual
or multiple node failures, self-healing. - Intelligence to anonymously react to information from their mesh
neighbors. - Rapid and coherent movement.
- Ability to avoid physical obstacles.
- Support for collaborative working.
Limitations
Historically, one of the common weaknesses of the massive robotics used in
industrial assembly applications has been the fragility of their cabling
systems. A typical robotic arm used in auto manufacturing includes six or more
joints; each of these joints flexes thousands of times per day, distorting the
cabling inside the arm and straining cable connection points.
For this reason, common causes of robotic downtime tend to include:
- Cable failure
- Wear and tear
- Loss of control
- System connectivity
- Other limitations related to scale, rigidity, and rapid movement.
In order to compensate for the robot’s inability to more naturally move, designers
tend to give them longer reach and faster operational cycle times, as a result
potentially making a tethered manufacturing robot dangerous for human workers.
Other complications could include:
- Inflation of factory size, maintenance cost, and retooling time.
- Added expenses related to reconfiguring fixed-position, limits on
innovation, and competitiveness.
Current View
[return to top of this
report]
Despite potential limitations, robotics remain an ever-growing segment of the
market in that they tend to be small, cheap, and mobile, with other distinct
advantages in manufacturing ability, scalability,
reliability, and – like virtually every area of emerging technology –
security.
"Real World" Manufacturing
Robots designed to participate in wireless mesh control systems typically
know only a few things: How to maintain stable motion (including, for
some, flight) and how to do simple tasks such as move up and down, forward
and back, turn actuators on and off, sense and record environmental data,
and the like. These robots are adroit but simple.
By contrast, the wireless mesh network is the conduit for intelligence
through which robots receive command and control information and upload
data for distribution or storage. This is a classic application of
“separation of concerns” strategy. By localizing process control
intelligence either in or above the mesh, it can be easily updated,
adapted, and repurposed. This type of architecture allows a swarm of
robots to have a different job from one deployment to the next and also
makes them jobsite independent.
For real world enterprise purposes, the key to creating a robust
swarm implementation comes down to having the ability to maintain
uninterruptible command, control, and communication with a robot swarm.
Wireless mesh networks are uniquely suited to addressing these
requirements.
Wireless Mesh Scalability for Robotic Control Apps
Most enterprise technology lives or dies on its ability to scale
gracefully. Wireless meshes have shown themselves to scale remarkably
well, owing to the fact that they are conceptually modular, fault
tolerant, and highly flexible. Wireless mesh can be embedded more or less
anywhere: In larger robots, in the environment, and even in vehicles or
aircraft that dynamically join and leave the wireless mesh as required.
Meshes can even join up with other meshes, making very large scale
networks possible. Currently, process-oriented industries employ three
levels of mesh controlled operation:
- Field Meshes operate at short range, sending up to a
few KB of data to neighboring mesh nodes fairly frequently. Field
mesh nodes generally use sensors to detect cues that invoke robot
behaviors like toggling power, triggering actuators, or movement of
physical controls. - Site Backbones operate within distances of a few
miles, update field mesh nodes’ behavior patterns and cues, and can
control operations of multiple meshes within their range. - Global Canopies use long range wireless
communications (such as cell networks or the Internet) to join site
backbones that are up to hundreds of miles distant from one another.
Unlike many radio technologies that have failed to scale well, wireless
meshes are globally consistent because all the component parts for a real
life implementation operate in the 2.4GHz wireless communications
frequency. By international convention, this part of the spectrum has
been set aside for industrial applications. Known as the Industrial,
Scientific, and Medical (ISM) radio band, it is not subject to change and
may be used without license. Standards including WiFi, Bluetooth,
WiMAX, and self-organizing/self-healing wireless mesh protocols like
ZigBee and Wireless HART all use the ISM band.
Security as a "Core"
Given these characteristics, it is safe to predict that the near future
will see large scale infrastructure systems (pipelines, railways,
electrical grids, and the like) becoming heavily reliant on wireless mesh
robotic surveillance and control. Aside from the obvious reliability and
efficiency advantages, wireless mesh robotics has something else to offer
these applications: Security.
From this technology’s inception, security was built into its core.
Wireless mesh uses highly compartmentalized, durable, redundant security
strategies and, like the rest of the design, these protections scale.
- Every data communicator device in a mesh has 128-bit encryption
capability. - The security manager and system manager services must each
authenticate a device before it can exchange data with mesh nodes. - Each data exchange point uses end-to-end verification to ensure data
accuracy and transport layer security. - Wireless devices cannot join a network without a key that allows the
rest of the network to authenticate them. - After a device joins the network, it is issued master, session, and
private keys, all of which have a defined lifespan and must be
revalidated periodically. - Data is transmitted spread spectrum and multiplexed. To reassemble a
data stream, an intruder would need to have both the spectrum spread
code and the frequency hopping pattern. - The mesh can blacklist individual wireless communication channels,
reducing the likelihood of invasion from another type of over-the-air
system.
Reliability
Beyond all this, there is the sheer reliability of mesh networks. Wireless mesh
technology has inherent
redundancy and self-recoverability. Mesh structure allows nodes to choose
transmission paths that avoid obstructions, to simulcast messages on more
than one channel, and to hop frequencies to avoid noise and congestion.
Outlook
[return to top of this
report]
Those who watched the opening ceremonies of the 2018 Pyeong Chang Olympics saw
swarms of small, flying, lighted robots elegantly form the Olympic Rings in
the night sky then melt away and reform as the dove of peace. It was
a watershed moment for swarming robotics, a once arcane technology. The moment was
instantly normalized by a performance that made use of market
ready systems and hardware. But while this was a showy and engaging
display, it wasn’t even remotely near the forefront of swarming
robotics research. To find that, you have to look in the opposite
direction of “big-as-the-sky.” Today, the real breakthroughs are at nano scale: Molecular robots constructed from cellular components that are
common to all higher life forms.
Molecular Robots Powered by DNA-Based
Operating Systems
In 2018, a Japanese research team lead by Akira Kakugo at
Hokkaiddo University1 demonstrated functional swarms of molecular robots.
These were built using a pair of proteins that play a role in cellular
metabolism; particles of DNA; and some small molecules called
isomers.
More specifically:
- The two types of proteins used were microtubules and
kinesin. Microtubules are the molecular robot’s
vehicles and the kinesin offers a system of roads that allow the
microtubules to travel where they want to go. - The DNA particle provides the molecular robot with on-board
“intelligence.” The technology is capable of the same kind of very simple
decision making that is found in the bottom of the technology stack of
any digital computer. - The isomer has properties that allow it to be an on / off
switch for the DNA “computer.”
Isomers are molecules that can rearrange their atoms in response to an
input. This is a pretty slippery concept, so here is a simplified example to
help ground this discussion.
Say we have an isomer it composed of the letters E-A-T. While the
isomer is in its EAT configuration, it switches on the molecular robot’s DNA
computer, which tells the robot vehicle to hunt for food. After the
robot gets a meal, the isomer is reset to have a new order: A-T-E. That isomer configuration switches off the DNA computer and causes the robot
to remain at rest until the cycle repeats. The isomer makes external
control of the molecular robot to be very precise.
To recap, a Hokkaido University molecular robot has a “vehicle” that can
move around on a transportation system; it has on-board intelligence and
decision making ability; and it has an on/off switch that can be remotely
activated. In the Hokkaido University experiments, the isomer “switch”
could be reset using visible or ultraviolet light. Kakugo’s team has
demonstrated that swarms of their nano scale robots can engage in relatively
sophisticated collaborative actions, including assembling themselves into
structures, doing coordinated motion and performing actions in reverse.
These robots are small. The microtubules synthesized by Kakugo’s team were 25 nanometers in diameter and
five micrometers in length. It is
probably not a coincidence that the smallest components of a silicon based
IC are 22 nanometers wide – just slightly smaller than the diameter of the
hollow microtubule “vehicle.” Molecular robot motion, flexibility and
freedom of travel are constrained only by the rigidity and shape of the kinesin surface on which it moves.
It is hard to overstate how transformational this advance is, as developments
are on the
threshold of practical applications of nanoscale manufacturing and biologic
construction. The near term implications for healthcare are staggering: An
injection of molecular robots could go on a “search and destroy mission”
targeting pathogens, cancer cells, etc. before disease symptoms even become
apparent. Illnesses like malaria, Ebola, AIDS and influenza could
become as distant a threat as goiter and scurvy are today. Preemptive
treatments could become game changers both in terms of efficacy and cost
savings:
- Swarms of molecular robots could perform periodic clean up and
restoration of artery walls, preventing atherosclerosis and similar
conditions. - Robotic intervention could interrupt the process of
addictions. - Why transplant organs if it is possible to fix and
revitalize the ones that are there?
Treatments for serious illness
that are costly, painful, uncertain and highly invasive may become obsolete.
And however great these prospects, they fall far short of “long game
thinking” when it comes to the future of swarming robotics.
Workforce Disruption on a Grand Scale
Cheap, reliable, flexible robot swarms are likely still a few years away from competing for every repetitive, dangerous or dirty job
that has sustained a vast, global population of low and limited skill
workers since the industrial revolution. This reality reaches far beyond
what we typically think of as “manufacturing,” but for purposes of
anticipating outcomes, the impact of mature swarming robotics technology
on manufacturing is as good a predictor of the future of work as one could
find. Every phase of futuristic large scale manufacturing employment will
be impacted.
Recommendations
[return to top of this
report]
The US Department of Commerce Manufacturing USA Program
has created and chartered a number of regional institutes2, each focused on a
unique, futuristic technology or manufacturing specialty. These institutes serve as R&D centers that take manufacturing
innovations from the proof of concept stage to the point of readiness for
commercial deployment. All locations are required to include US workforce
development education as an integral part of their overall program. Work
includes advanced optics, intelligent textiles, sci-fi worthy
military devices, 3D printing of human organs, spreading / accelerating the US “Maker Movement,” lightweighting, and
conducting other manufacturing innovations.
Web Links
[return to top of this
report]
ANTS Swarm Intelligence Conference:
http://www.swarm-intelligence.eu/ants2018/
Department of Commerce Manufacturing USA Program:
https://www.manufacturing.gov/partners/department-commerce
Swarm Robotics Project of the EU Future and Emerging Technologies
Program: http://www.swarm-bots.org/
References
1Hess, Henry, Daisuke Inoue, Arif Md. Rashedul Kabir, Akira
Kakugo, Jakia Jannat Keya, Akinori Kuzuya, and Kazuki Sada. "Control of swarming
of molecular robots." Nature.com. August 6, 2018.
2"How to Engage with DOD Institutes." Manufacturing USA.
Accessed March 2020.
[return to top of this
report]