During the 1980s, the information technology revolution transformed the global economy, requiring whole sectors to re-model how they operated and enabling organisations to change how they delivered products and services. Today, a similar technical revolution is emerging that once again presents organisations with the opportunity to reshape how they do business. This advancement in technology is the ability to remotely capture data using unmanned piloted systems and use machines to think for us, using the power of deep machine learning. These two technologies will change how businesses make decisions. Combining these two new ways of working provides infrastructure operators and asset managers with really powerful insight into how structures, earthworks and equipment are performing, enabling critical points of failure to be identified and thereby instigating earlier interventions to stop outages to service function – in effect making infrastructure intelligent. The need to survey An example of how data and intelligent infrastructure technology has a wider potential for the rail industry was covered in two recent articles, ‘The Internet of Things’ (issue 138, April 2016) and ‘Control and communications – asset management’ (issue 143, September 2016). Both articles explored how capturing data and processing more intelligently empowers rail operators and asset owners to make smarter decisions. This is entirely feasible considering that the UK rail network is made up of over 20,000 miles of track and 40,000 structures, all requiring regular inspection, monitoring and maintenance. Whilst advancements in some areas have been made, this work is still predominantly completed through traditional surveying and inspection methods. But are these sustainable to support a growing railway of the future? Conventional inspection and surveying is often labour intensive and time consuming, requiring personnel to access structures within hazardous or difficult-to-reach environments. Equally, it is sometimes impossible to deploy personnel to certain locations due to the high-risks involved or to physical environmental barriers, meaning the data that is needed to make decisions on critical infrastructure is simply unavailable. Where personnel have to be deployed, strict safety procedures have to be implemented to protect the workforce, often resulting in disruption to train operations and the knock- on impact to the passenger. Accordingly, asset owners and operators need to balance the operational necessity of maintaining infrastructure with reducing disruption to the network. Dealing with data This is where Unmanned Aerial Systems (UAS) or drone-powered solutions can add value. In September 2014 (issue 119), Rail Engineer reported “Rail Survey technology reaches new heights”, so the use of drones is not new to the rail industry with a number of organisations already using this technology as part of their workflow. An example of this was covered in Mark Phillips’ article “Cliff terrain surveys using UAVs” (issue 139, May 2016). But using unmanned aerial systems can also support a longer-term strategy of reducing ‘boots on ballast’ whilst offering infrastructure managers the ability to capture data from previously inaccessible areas not possible using traditional methods. This opens the possibility of being able to gather data easily and more frequently. However, simply having the ability to increase inspections of multiple assets starts to present a different problem. The more frequently a structure is inspected, the more data is captured. This ultimately needs analysing, which increases the demand on specialists to review the data – and so the cycle continues. But current estimates are that less than 0.5 per cent of all data collected (not just in the rail sector) is ever analysed, so remotely capturing data on its own is clearly not the answer. This is where emerging data processing techniques, often referred to as cognitive learning, can add real value. Using cognitive or ‘deep machine’ learning, these advanced processing techniques can autonomously review data, enabling images to be used more intelligently. This means that asset owners, managers and operators can make better-informed business decisions, enabling the dynamic use of resources more safely, quickly and efficiently, ultimately optimising whole-life cost whilst reducing the impact on passengers and rail freight operators. But what needs to be considered when integrating remotely piloted systems into a business’s workflow? Introducing UAS technology into the rail industry, as with any new system of work, requires necessary safety concerns to be addressed. As with any emerging technology, the maturing UAS market now offers an increasing range of systems and operators. But with this choice comes the need to understand the capabilities and limitations each of these various options offer. Specialist equipment Whilst a large number of ‘drones’ are available on the market, in reality there are only a limited number of systems that are capable of operating successfully in challenging commercial environments. One UAS that has been designed purely with commercial operations in mind is the Altura Zenith ATX8 from Aerialtronics – a global manufacturer of commercially designed remotely piloted systems. The ATX8’s strength is its reliability, stability and versatility, having been built specifically for the commercial market. It is able to accommodate a range of sensors or cameras which can be changed simply and quickly in the field. A flight time of 25 minutes, coupled with its ability to operate in 14m/s wind and carry a payload of 2.9kg, gives the ATX8 significant flexibility not available with a number of other models available on the market. Also a purpose-built UAS gives confidence to regulators, such as the Civil Aviation Authority (CAA), who are concerned about the stability and safety of operating such a device in challenging or high risk areas. This is important, particularly since railways are a predominant feature of the urban landscape, as regulators need to grant additional permissions to operators using UASs in congested areas, beyond the normal agreements granted as part of the standard Permissions for Commercial Operations (PfCO). This confidence has resulted in telecoms and power distribution companies remodelling their workflows to use UASs, saving significant time and operating costs when inspecting mobile communication towers and high-voltage overhead powerlines. One telecoms company doing this is T-Mobile, which recognised that inspections of cell tower masts that could take up to seven days using traditional methods (for example using a team of technicians in cherry pickers) now take a third of the time using a UAS. The Netherlands-based energy infrastructure management company Joulz also uses UAS technology – an ATX8 assists in the management of high-voltage overhead lines. Maintaining overhead powerlines involves examining pylons, inspecting insulators, and detecting thermographic problems. Inspection is predominantly dependant on technicians climbing structures to access power lines, or by using helicopters. Unmanned aerial systems provide a safer, cheaper, and easily deployable alternative. A camera that produces high resolution and thermal images enables the UAS to record and transmit live footage both to engineers on the ground and to experts located elsewhere, making the whole process more efficient. Another example is a wind turbine installed in a remote area of Scotland. Its blades can be inspected by UAS. If a critical issue is identified during that inspection, the image can be immediately transmitted to the manufacturer in Sweden where the data can be reviewed by a specialist team and specific advice given concerning appropriate remedial action. Intelligent decisions But the needs of infrastructure owners and asset managers to make effective decisions can be improved even further through combining remote data capture technology with enhanced data processing, or cognitive learning capability. In simple terms, this uses computers that learn and understand what we, as humans, are looking for through the application of visual recognition. An example of this would be identifying corroded structures, loose or damaged cabling and equipment and for computer processing to automatically review images during future inspections to identify whether the issue is getting worse. Therefore, combining visual recognition and remote data capture technologies enables teams to inspect critical infrastructure by deploying UASs to gain an immediate 360-degree, high resolution overview of a structure. This data can then be sent from the site for immediate cognitive processing and near real-time analysis. This cognitive analysis of images over an ongoing period can result in specific areas of concern being identified through computer processes that are constantly learning – building points of reference that understand what to look for in future data. Using this combination of technology enables asset owners to continue to provide safe operations across the rail network through supporting engineering teams to have an enhanced evidence base to make informed decisions and prioritise remedial action. Capturing images safely and dynamically allows efficient analysis of data whilst targeting the use of specialist human resources and reducing exposure of the workforce to undue hazards. Longer term, the data provides information that will increase asset owners’ and users’ confidence in the integrity of critical infrastructure and therefore underpin decision- making, including future investment strategies. Over the next decade, combining remote data capture techniques with powerful cognitive computer learning will bring significant benefits to the rail sector which will be felt by asset owners, infrastructure managers and, ultimately, passengers. Furthermore, this powerful analysis can be enhanced further when UAS-captured data and cognitive learning are linked with Building Information Modelling (BIM) system data. This will provide clients with a range of interests in critical infrastructure to build long- term, evidenced-based organisational memory about the structures they rely on. In turn, this new technology will also contribute to reducing the need for the rail workforce to be exposed to hazardous environments, as well as reduce downtime from infrastructure outages, structural failures, or unscheduled work. Over time, the costs to asset owners, operators and users will reduce due to efficient inspection processes and improved data supporting decision making about planned maintenance. This technology also supports more targeted responses when assessing the damage to infrastructure following disasters, for example major flooding or extreme weather events, to which the UK is becoming exposed.