In June 2019, experts in Customs, trade, transport, logistics and technology gathered in Baku, Azerbaijan, to discuss matters relating to the various technologies used to manage flows of goods, people and means of conveyance across borders, with a focus on both technological products, services and tools that are primarily physical and those that derive their primary value from data.

This newly-launched annual event, which merged the WCO’s long-running IT Conference & Exhibition with its newer Technology & Innovation Forum, has the same broad objective, namely a platform to inform Customs on existing tools and services, to discuss implementation challenges and costs, and to raise awareness about the latest trends and future developments.

The process of “selling” specific technologies usually means selling a vision, convincing people to adhere to that vision, and then informing them about the technologies that can be used to implement the vision. Participants, therefore, did not only get the opportunity to test and understand how solutions and equipment work, they were also asked to think about a vision and how to adapt their business models accordingly, including anticipating the future.

Below are just some of the ideas gleaned from the event.

Customs has a lot of information at its disposal. The first information they collect is pre-arrival declaration information related to shippers, consignees, values and commodities, i.e. typical manifest information. Additional information is collected at arrival, through the Customs declaration and through various sensors, including non-intrusive inspection (NII) systems. If Customs or another agency decides to check the cargo, the inspection results will also be added to this pot of information.

However, Customs does not always use this data in a suitably combined manner. For example, Single Window systems may not include inspection data. Therefore, a time of transition will be faced in looking for solutions to combine all this data. Interestingly, the NII industry has been creating new products that provide richer data sets, with some systems now combining various X-ray technologies, displaying the generated data set on one user interface.

The X-ray image file itself is enriched with additional external data to generate a more informed outcome. NII industry leaders have developed platforms, which not only enable images to be adjudicated remotely, but also allow for the integration of data, including data stored in Single Window systems, as well as data from inspections, images, radiation monitors, biometrics and weight scale devices, to name a few.

Single Window systems are now seen as the main aggregator tool for collating data, such as that from the invoice, the bill of lading or the packing list, among others, received from an importer or exporter’s enterprise resource planning (ERP) system to prepare regulatory documents. One of the systems presented at the conference had the following key functionalities:

• It can notify a trader of any additional data that is required to fulfil a submission.
• It can apply risk rules to flag shipments that display known risk characteristics, where discrepancies exist between information on the manifest and the declaration.
• It uses artificial intelligence (AI) to identify unusual “behaviour” in declarations, including correlations or similarities with previous shipments that were mis-declared.
• It can run a network analysis and identify associations with blacklisted individuals, entities or locations.

Advanced risk management tools are also becoming more and more sophisticated. One of those dedicated to monitoring means of conveyance is able to mine and display an incredible amount of information, such as, for the maritime sector, current positions of vessels, history and previous movements, vessel ownership, and the nature of the cargo.

A lot of information is still available on paper and this information needs to be extracted and the document reconstructed in a digital format such as Excel or Word. This is now an easy process, and can be done using a phone camera, for example.

Even in this era of “big data,” one might say “you ain’t seen nothing yet, just wait until the Internet of Things takes hold and really expands.” The term “Internet of Things” (IoT) is used to describe an environment in which sensing objects send data to an application, data that can then be used to create situation awareness and enable applications, machines and human users to better understand their surrounding environments. The understanding of a situation, or the context, potentially enables applications or humans to make more intelligent decisions.

IoT is already everywhere, and is expected to grow. Connected devices are widely used for supply chain management to track assets, monitor cargo condition, and control its environment such as the temperature. It enables a log of wood, for example, to be followed along its transformation chain from the place where it was harvested to the place where it will be sold.

Connected devices are also used for fleet management to gather data on the locations and operations of vehicles in real time, as well as to control sensitive parts such as engines, to predict maintenance or optimize repair operations. A spare part of a plane can be ordered before it arrives, and a truck system failure can be identified and actions automatically initiated according to pre-determined workflows to send help or a new truck, including informing the place of destination and authorities of expected delays if necessary.

Thanks to visual recognition tools, one can teach a machine to recognize goods, such as arms or bad grains. Add to this the capacity to read and translate text, and one gets a technology that can see what the object is and read what is on it.

This is how a fridge can identify what it contains. Some scanners today take images and apply this technology to recognize items: at an airport for example, or when cargo is loaded. If one combines the image and other information related to the content, origin, or owner of the cargo, one can determine risk and take action according to the level of risk.

Aggregating data into one single platform will enable one to move towards machine learning. Indeed, one needs a lot of data to leverage the technology, which is directly related to pattern recognition and computational learning.

The development of algorithms to discover anomalous behaviours in real-world transactions or processes requires access to information related to the goods, the image if any, and the final inspection decision as well as feedback.

Intelligent systems can also predict the arrival or positions of ships, by combining information related to the reputation of the shipping line, encounters at sea, changes in the speed of a route, the bill of lading, or weather conditions.

To do such prediction requires a lot of capacity to store and process data, especially if one wants to predict the positions and the arrival of fast-moving objects such as planes, as one will need multiple updates per second, per plane.

The use of machine learning systems, in terms of hardware and software, was until recently out of reach for most, but the rise of cheaper data storage, cloud and non-cloud, that makes the same massive data sets available from the same source, changed the game. Moreover, the power that one now has to process data has also improved, with solutions such as Hadoop boosting the processing speed of data analysis applications.

In building risk management frameworks, one wants to combine all sorts of data from different sources. Thus, the “data lake” is a new approach – a central repository for storing large amounts of raw data that can be analysed when needed.

It is typically used to store data that is generated from high-velocity, high-volume sources in a constant stream – such as IoT – and when an organization or entity needs a high level of flexibility in terms of how the data will be used. Unlike a “data warehouse,” a data lake has no constraints in terms of data type, which can be structured, unstructured, or even semi-structured.

Internet connectivity is a two-way street. With connected devices becoming a gateway to homes, workplaces and sensitive data, they also become targets for attacks. The industry’s concern about the security of connected devices has been late in coming, and one of the problems with such devices is that most have no security capabilities whatsoever.

However, a technology provider has developed a solution, combining a secure microcontroller (the brain of the device), operating system and cloud service to improve the security of IoT devices.

A unified X-ray file format for NII devices – codenamed the Unified File Format (UFF) – has been developed, in cooperation with the WCO, by four NII vendors: L3, Nuctech, Rapiscan Systems AS&E, and Smiths Detection. The first version of the standard is called UFF 2.0.

It will be further developed depending on the rate of its adoption by Customs administrations who have been invited to require their suppliers to deploy the UFF on equipment in use, if possible, and to add the specifications of the UFF 2.0 in their tender documents for the procurement of NII systems.

Each of the four vendors participating in the UFF initiative developed a UFF 2.0 converter and a UFF 2.0 viewer – software enabling the conversion of native NII images to the UFF, and respectively the viewing of UFF images with all the tools and functions available in the original image viewing/processing software.

The UFF will significantly facilitate the interoperability of NII equipment supplied by different manufacturers, as well as the exchange of NII outputs within and between Customs administrations. The UFF will also be instrumental in developing the large databases or libraries of images that are, inter alia, necessary to train analysts or machines equipped with an automated detection application.

5G is an umbrella term used to categorize the next generation of wireless communication, offering networks that are 100 times faster (to move more data), connect a lot more devices at once (for sensors and smart devices), and feature five times lower latency (to be more responsive). In the immediate future, this standard will mean clearer video conferencing, faster downloads, and never losing service in a large crowd.

More importantly, further down the road, 5G has the potential to revolutionize aspects of life by supporting more data-intensive activities, such as the IoT. 5G will provide the infrastructure for handling thousands of devices simultaneously, from mobile phones to equipment sensors, video cameras to smart street lights, and interconnected traffic lights to autonomous vehicles.

Data processing in an environment such as airports, where time is a constraint, will benefit greatly from it. It will also be crucial to move and process data of significant size such as scanned images. Finally, it will be a key foundation point if one is going to use technology such as blockchain and AI.

Integrating real-world data into the Web and providing web-based interactions with IoT resources was also discussed under the umbrella term “Web of Things.” The idea is to connect the sensor, actuator and other devices to the World Wide Web, enabling web-based access, tasking, and alerting.

Connecting this diverse data, however, presents an ever-growing challenge, and the greater potential of IoT will not be achieved without standards-based vocabularies and reference data that can be mapped and connected to each other.

This is where technologies based on the Resource Description Framework come in. Their strength is in connectivity. They enable diverse data from different applications to be brought together easily. With the semantics of data directly accessible to applications, each can interpret merged data according to its needs.

The current level of collaboration is still too low to get all the effectiveness needed to manage risks properly. Hopefully, this will change, which means that platforms being developed need to be able to connect to each other. In such a connected landscape, all actors would have access and be joined up to one single source of “truth” through a single domain, where they would gather information and harvest it.

Machine learning needs to be embedded in the workflow. Let’s take the example of two machines dotted with learning capacities working separately, one managed by Customs and the other by another national agency or foreign country. One will need a sort of “supra machine” to learn from both, and blend the intelligence build of each machine, without the need to transfer any data.

AI-as-a-Service is a concept that comes from the medical industry where images are not read at the location where an X-ray is taken. It implies sharing information with a third-party vendor. For example, when it comes to cargo or baggage, one country could upload images and related data onto a server. A service provider will then process the data using algorithms tamed to read specific anomalies and render a decision for a receiving administration before the cargo or baggage arrives.

It would be especially relevant when dealing with air passenger bags, which are automatically scanned: an AI machine would scan the scanned images and process data while the plane is in the air. At destination, an administration would receive the outputs of the analysis and the list of bags flagged.

Most technology solutions are energy-intensive. When considering their deployment, one should first ensure that energy is available, and that it is produced in a green and renewable way with a view to minimizing the repercussion of the investment on the environment.

Research on the impact of technology mainly focused on blockchain and more especially Bitcoin’s environmental footprint. In 2018 alone, the running of Bitcoin mining around the world took as much energy as that necessary to fuel Ireland’s total electricity spend in the same year. One single bitcoin transaction requires as much energy as that needed to power 31 United States households in a day. The number goes up to 1 million transactions, in a day!

However, some argue that, if Bitcoin requires a digital-labour intensive process with excessive energy demands, blockchain does not necessarily. In any case, the energy systems that are fueling the transition towards advanced technologies should make use of the right energy sources.

It is used in applications to compile albums of people who hang out together, at airports to verify who you are, to unlock your mobile phone or confirm your identity for a bank transfer, to know who is at your door, to spot missing persons, or find people in crowds and on city streets. For example, if a suspect is picked up, police officers can upload the person’s mugshot and search CCTV footage to potentially trace the suspect’s movements back to the scene of a crime.

Advertisers are also in on the act. Thanks to facial recognition, billboards can now serve up ads based on an estimate of your sex, age and mood. But how did it get everywhere? Advances in three technical fields have played a major part: big data, deep convolutional neural networks, and powerful graphics processing units, or GPUs. Customs administrations use facial recognition mainly at airports to scan passengers, quickly identifying those that need to be controlled.

It is also used by the transport industry to monitor drivers. Cameras mounted on the dash monitor if the driver’s head starts to tilt or his eyes start to close for too long, or if his head position is not right or his eye line is not straight ahead, or if his blink rate has increased, and even how many occupants are in the vehicle. The system will then decide if any intervention is needed, which could be in the form of an audible warning or a light, a vibration of the seat or steering wheel, and possibly even a slight take-over of the vehicle controls – slowing the car down or moving lane to a lower speed.

Trade facilitation is still very government-driven, but the increased number of pilots leveraging blockchain technology to manage cross-border transactions show that the logistic and transport sector is ready to be more active in the development of solutions, providing more transparency about data upstream in the supply chain as well as real-time visibility across the end-to-end journey of a shipment, with the aim of increasing efficiency and reducing operational costs.

The notion of “Internet of Logistics” (IoL) is also much talked about. The initiative, driven by air and maritime transport companies, aims to transform peer-to-peer message-based communication towards multi-party data sharing to increase shipment visibility.

A blockchain is a type of distributed ledger technology (DLT) with a specific set of features, including a shared database (log of records) shared by means of blocks that form a chain. As various and sometimes competing blockchain platforms develop around the world, they must also interact.

Integration between different platforms will be necessary to maximize their benefits. Integration means that two or more DLT platforms can transfer an asset with the confidence that its uniqueness and state are kept consistent. It would also allow a consistent state of a data element on two or more DLT systems to be maintained simultaneously.

Administrations willing to harness the power of technology have to transform their business and change how they operate and deliver value to customers. Digital transformation is all about business transformation.

A roadmap to harness the true capabilities of technology needs to be developed, which includes a cloud policy, a data sharing plan, workflows explaining what needs to be done, and mapping how the technology will be used to do so, as well as the value proposition for customers.

The transformation of transport will first be characterized by the growing use of the autonomous truck, a sector in many countries desperate for drivers and still suffering from inefficient transport systems. In Africa, the truck will not go totally driverless, but the technology will be used to support drivers, especially with long-distance hauling or on dangerous roads where they cannot stop.

Drone usage will also spread, especially for the delivery of cargo to remote areas with no infrastructure to accommodate manned cargo planes or with inadequate or poor road infrastructure, resulting in low road utilization. Africa is leading the way in terms of drone usage with the world’s first commercial drone delivery service having begun operations from a hill in the middle of Rwanda.

Zipline, a San Francisco-based robotics company, delivers blood by drone to almost half of all Rwanda’s blood transfusion centres, with orders being made online, or by text, phone or WhatsApp. The South African National Blood Service will shortly be using drone technology for the same purpose. Kenya’s Astral Aviation uses a 2,000 kg payload drone that can travel up to 1,200 km, solving the logistics challenge of transporting cargo to remote areas.

The 2020 WCO technology event will be held in the first half of June 2020 in Bali, Indonesia. Keep checking the Events Section of the WCO website for further details. If you missed this event, the WCO looks forward to welcoming speakers, delegates and exhibitors alike to the next one.

More information
www.wcoomd.org