Cross-Industry Whitepaper Series: Empowering Our Connected World
New technologies and initiatives in the field of robotics
Key topics we will discuss in this paper include:
Role of AI in helping robotics perform better
How the Industrial Internet of Things make factories and
supply chains more efficient
ROBOTS
BREAK OUT
OF THE FACTORY:
NON-CONSUMER
ROBOTICS
APPLICATIONS
TAKE OFF
Impact of open source and cloud computing on robotics costs
Drones and the world of robotics
Breaking barriers with advanced mobile communications networks
INTRODUCTION
Today, there are established uses of robots across multiple
manufacturing industries. These include automotive,
aerospace, electricals/electronics and metals/machinery,
food/beverages and pharmaceuticals. Outside
manufacturing, robots are also used in agriculture,
defense, and for moving things around in many sectors.
Robot tasks include welding, cutting, materials handling,
assembly, cleaning, spraying, and so on.
Statistics compiled by the International Robotics
Federation indicate that 254,000 industrial robots were
shipped in 2015, and the organisation forecasts that by
2019, there will be 2.6 million industrial robots deployed
worldwide. Sales are predicted to grow at a CAGR of 13%
over the next three years. Robotsare being installed in
industrial settings all over the world.
Supporting this growth is rising research and development
spend: analysts Technavio estimated in July 2016 that
robotics R&D spend would increase at 17% per annum up
to 2020, with most R&D spend by robot OEMs .
Traditional industrial robots – mechanical devices designed
to replace or enhance the ability of a human to perform
typically repetitive tasks in specific industrial settings – are
already the most well-developed commercially.
But the industrial robot market is at the start of a wave of
development where breakthroughs in other technology
areas will increase the scope of what industrial robots can
do and lower the cost of robot systems. These include:
systems (“co-bots”) and integration with the wider
supply chain
exciting changes in industrial robotics).
In factories, robots for performing manufacturing tasks
such as welding can work closely with those moving
materials, components and manufactured items around.
Robots for logistics applications also exist within many
other industries such as wholesale/retail warehouses,
libraries, pharmacies in hospitals, and baggage handling
in airports.
(Note – in this paper we exclude autonomous vehicles for the
public highway from our definition of industrial/logistics robots,
though we include autonomous guided vehicles in specific
non-highway settings.)
Robotics is a very active area of technology right now.
There have been long-standing uses for robots since the
first industrial robot was installed by General Motors in
Trenton, New Jersey, USA, in 1959 .
1 Unimate: The First Industrial Robot – robotics/joseph-engelberger/unimate.cfm
2 ifr/industrial-robots/statistics/
3 ifr/news/ifr-press-release/world-robotics-report-2016-832/
4 technavio/report/global-robotics-r-and-d-spending-robotics-marketutm_source=T3&utm_medium=BW&utm_campaign=Media
5 https://ft/content/d8d80f32-3874-11e6-a780-b48ed7b6126f
2,3
4It’s all about economics. The costs of the systems are
falling and the capabilities are rising, we have at least
15 to 20 years where robotics will continue to expand
- Hal Sirkin of the Boston Consulting Group, June 20165
2.9
MILLION ROBOTS
By 2019, there will be at least 2.6 million
industrial robots deployed worldwide
13%
Sales are predicted to grow at a
CAGR of 13% over the next three years
254
THOUSAND ROBOTS
254,000 industrial robots were shipped in
2015
17%
PER ANNUM
R&D spend would increase at 17% per
annum up to 2020KEY DATA
NEW TECHNOLOGIES AND INITIATIVES ARE EXPANDING
THE SCOPE OF ROBOTICS IN AND BEYOND THE FACTORY
While uses for robots can be identified in very many sectors,
take-up is not universal in those sectors.
In part this is related to the relative costs of specifying,
acquiring, deploying and operating robotic systems on the
one hand, and local cost of human labour.
Robotics systems have traditionally been an expensive
upfront cost. For some applications, the costs of the “end
effector” (the application-specific device that goes on the
end of the robot arm) and “peripherals” (for instance
transport systems on which robots may run, or
communications networks) can significantly add to the cost.
In most cases, programming the robot to do what is
required and to respond appropriately in different scenarios
can be a significant cost. These issues are now being
addressed by developments in several areas of technology.AI, MACHINE LEARNING AND DEEP LEARNING
ENABLE ROBOTS TO PERFORM BETTER
The idea of teaching robots to do a task, rather than programming
a control system to achieve the same task, has been developed
over the last ten years or so, as the academic field of Artificial
Intelligence (AI) has become commercialized.
Increases in the power and affordability of computer
processing (coupled with fast networks) have been so great
that consumer applications such as Google Now, Apple’s Siri
and Microsoft’s Cortana can understand questions posed by
users, find the relevant answers, and respond in natural
language within milliseconds.
Within the broad AI domain, machine learning describes
the use of algorithms (of various kinds) to help computers
to interpret data and make decisions based on the data.
They can be “trained” to understand when their decisions
are right or wrong so that their decisions get better over
time. Machine vision systems and email spam filters are
among applications that have used this approach. Deep
learning refers to the use of multi-layered artificial neural
networks that enable the training to be carried out on a
huge scale, with the result that decisions are very much
better. An example is the Go-playing robot Alpha-Go that
defeated world champion Lee Se-dol in 2016, or more
usefully, systems that can examine MRI scans and identify
indicators for illnesses.
Using machine or deep learning, a robot can become
better able to complete a task, or to undertake a new one,
through an improved awareness of its environment and
the context of the task. These approaches will also
reduce the need – and cost – to program robots for each
new task. This in turn creates the possibility of more
flexible industrial robots able to cope with changes in
factory configurations and shorter production runs, and
capable of optimizing the processes that they are
required to perform.
Access to the compute power required for machine and
deep learning is greatly enhanced through high-speed
networks and the use of cloud resources.
Hello!。。。以上简介无排版格式，详细内容请下载查看