Use Case Series: 2. Transport & Logistics’ Advanced Analytics & AI Opportunities

We are currently in another transformational period in human history where the digital revolution is reshaping different aspects of our modern world, just the way the industrial and agricultural revolution did years ago. At the forefront of the transformation is AI. 

AI is leaving indelible marks on every industry, the transport and logistics industry included. DHL in its Logistics Trends Radar report identified AI and ML as some of the important technologies driving innovations in logistics. With its ability to analyze and make deductions from massive data, as well as automate the operational process, AI is enabling business’s supply chain to enjoy a competitive advantage through a transformation of their logistics angle.

Let us examine five areas where AI is impacting the transport and logistics industry.

1. AI in Logistics Administration

The business world is increasingly becoming complex and competitive. And amidst all these, clients still exert unprecedented pressure by demanding higher service levels at lower costs. At the same time, the internal functions of transport and logistics companies like finance, legal, IT and human resources are under the pressure of dealing with detail-oriented repetitive tasks. Here, we examine what time-saving, cost-reducing and productivity-boosting opportunities AI has to offer in the back office processes.

Data Analysis

A lot of companies now have access to big data which helps to provide information on different aspects of the supply chain. However, big data can only prove beneficial when it can be interpreted easily and quickly. Having a team of analysts working on the information from big data can’t leverage the benefits of big data to its maximum. The ability to get insights from this real-time data can be a complex and time-consuming process, since the data collected can be so varied and hard to understand even with the analytical algorithms developed to support this analysis. 

This is where AI comes in; it uses machine learning, deep learning, and natural language processing to automate the creation of new algorithms, with minimal human intervention. With this, the information from big data in the supply chain can be better streamlined and automated to allow logistics operators make quick decisions and avoid wasting resources on a team that’d have to mine, then interprets large volume of information. So, while a lot of companies might already be leveraging big data, without AI technology, they are at the risk of being data rich and insight poor.

Cognitive Automation

AI can help eliminate routine and mundane tasks to allow your workforce focus on more meaningful and impactful work. Here, AI is used alongside Robotic Process Automation (RPA) to automate business processes. They both help to cut down on the tasks of the logistics workforce by introducing software robots that are integrated into IT systems to automate tasks. AI is quite different from RPA in the sense that RPA works based on set rules to automate tasks like access and copy data, fill web forms and make these data sources available to AI. Then, AI as a self-learning system learns, judges, interacts and improves processes.

This way, repetitive and mundane tasks like file checking, collecting, copying, and other detail-oriented tasks that are being previously handled by human agents can then be passed on to robots. Humans can then focus on other complex and more important tasks like interpretation and decision making.

Detection of Financial Anomaly

Transport and logistics companies often require the service of third-party vendors such as subcontracted staffs or common carriers to ensure smooth operations. These several third-party providers translate to a sea of invoices annually for your accounting team to process. AI can help in this situation with the use of natural language processing that is capable of extracting vital information like billing amount, addresses and account information from the millions of unstructured invoice form you receive.

Once the data has been properly classified with the use of natural language processing, then an RPA bot can then work on the data collected using accounting software and process the customer’s order, payment and deliver a confirmation email as well.

Contract Management

Logistics service providers that operate globally have a fleet of vehicles as well as a network of facilities they manage worldwide. Therefore, they are faced with processing and managing a lot of contracts. Here, AI technology – natural language processing can be used to simplify this process. A company that has cashed in on this is Leverton which uses AI on its platform to classify contractual clauses, signature portions and policy-relevant sections in a fraction of the time it’d take a team of humans to do it. And, when a human is brought into the loop, these already classified legalese contracts can be processed even faster. 

Customer’s Information

You’d agree with me that keeping current and complete contact information is vital for the successful delivery of goods. However, large enterprises are increasingly faced with the challenges of having up to date customer information. According to Deloitte, 27% of email addresses and phone numbers stored in digital contact application are no more useful. In most cases, logistics companies do have a team of analysts who oversee tasks like duplicate entries, eliminations, removal of old contracts and data format standardization, and which of course takes up a lot of time. 

AI can help to provide input management services to reduce the burden on your workforce. It services here includes but is not limited to scanning letters, entering invoices into accounting software, uploading spreadsheet and presentation in the cloud, preparing and processing of customer’s contact information and ensuring it is complete, correct and consistent with the regional and global address format 34. CircleBack, an American startup have developed an AI system that can process billions of data points to determine the completeness and correctness of contact information.

2. Predictive Logistics

The logistics world is characterized by uncertainty and volatility. And, AI has the potential for providing network and process prediction to enable logistics companies to move from reactive to proactive using predictive intelligence.

Predictive Network Management

AI in-network level prediction can provide a significant improvement on logistics operation. Let’s take air freight for instance; air freights represents 35% of global trade in terms of value but only 1% in terms of tonnage. A look at the lanes and networks used by air freight reveals that they are planned using historical data and the expertise of industry professionals and there are still cases of unexpected delay. 

A machine learning based tool developed by DHL can analyze 58 different parameters of internal data to predict air freight’s transit time delays for a given lane, to enable proactive mitigation. In addition to this, the machine learning model can identify the top factors like temporal or operational factors influencing delay in shipment. With this, air freight carriers can remove subjective guesswork regarding the timing or choice of airline for their shipment transportation.

Predictive Demand

Predictive AI doesn’t just help logistics operators with network management; it can also help predict demands of goods, to enable them plan their shipment capacity ahead of time to meet up with demands. Let me cite the instance of the fidget spinner toy boom of 2017. Videos of the 3-paddled shape toy used for tricks by teenagers began trending on YouTube in February 2017. Within a few months, there was a boom in demand for the spinning toys as it unexpectedly sold an estimated 50 million unit in such a short period.

This boom led to a large sudden and unanticipated rise in supply chain demand, causing toy merchants to reject the normal ocean shipping of their goods, and leading the air freight and express networks inundated. 

Now, what changes could AI have made in this case?

AI can go through unstructured data like trending YouTube videos, trending social media conversations and hashtags, as well as browsing data to identify a qualitative rise of interest in a topic and predict which fad could boom, just like the fidget spinners did. This lead time can provide significant advantage to both merchants and logistics companies that could be faced with an unexpected spike in demand.

An example of this is how world logistics giant DHL, uses its Global Trade Barometer tool which uses AI advanced statistical modeling and operational logistics data to indicate current and future development of global trade. The model takes a bottom-up approach to evaluate 240 million variables from 7 countries that represent 75% of global trade (China, Germany, Great Britain, India, Japan, South Korea and the USA) to provide an effective three-month outlook for global trade.

Predictive Risk Management

Supply chain leaders in different industries, including the technology and manufacturing sector, conducts business with thousands of worldwide suppliers. As a result, different suppliers’ problems ranging from material shortages to legal investigations could crop up, leading to disruptions in the supply chain. Therefore, having a predictive risk management system in place can help ensure supply chain continuity.

The DHL Disruption360 supply watch module which uses Machine Language and Natural Language Processing help identify and mitigate supplier-side risk. The Resilience360 supply watch module examines the content as well as the context of over 8 million posts from over 300,000 online sources, to identify the sentiments of conversation and identify indicators of risk ahead of time to enable supply chain managers to take corrective actions and avoid disruptions.

Traffic Management Operations

It is quite interesting to know that AI solutions have been applied in resolving traffic optimization and control problem. Some of the use cases of AI in traffic management operations include the prediction and detection of traffic conditions and accidents. This feature has been achieved with the use of traffic sensors that are converted into intelligent agent using cameras. An example of this is Surtrac by Rapid Flow Technology. In June 2012, a pilot of the Surtrac system was installed by Rapid Flow in Pittsburg. The solution was applied to nine traffic signals in three major roads. During the trial, it helped cut down on travel times by more than 25% and wait time was cut down by 40%

Intelligent Route Optimization

The ability to optimize route is very important for logistics providers to efficiently transport shipments. Most often, logistics operators have both explicit and implicit knowledge of cities and their characteristics. However, with new customer’s demand such as instant delivery and ad-hoc pickup on the rise, it creates new challenges for intelligent route optimization. 

A lot of logistics company struggle with route optimization. In fact, this problem has been in existence as way back as 1930 in the form of the ‘travelling salesman problem.’ The traveling salesman problem sought to ensure optimal use of time and resources by finding the shortest routes connecting cities. Previous approaches like heuristics and optimization solvers aimed at addressing the traveling salesman problem all had the shortcoming of being unable to learn. They don’t gather experience from the problem that was solved. 

AI as a self-learning system can help with this problem.

AI can work in tandem with city infrastructures such as satellite and digital maps as well as traffic patterns to improve the routing of the truck drivers on delivery runs. DigitalGlobe, one of the leading Earth Observation (EO) in the world provides images of the earth surfaces to Uber. The satellite Imagery Company can provide details like lane information, street scale changes to traffic patterns to create valuable insights to navigating routes for transportation.

3. AI in Autonomous Logistics

The use of autonomous vehicles is now in place to improve the supply chain. For autonomous vehicles to outperform human driving capability, they need to be able to perceive changes in their environment – a task unachievable without AI. The ability of developers to program autonomous vehicles to react to different driving scenarios in the world will be greatly dependent on deep learning algorithms. With deep learning, they can develop autonomous vehicles that can improve their capability to function in new surroundings and be able to identify cars, road signs, traffic signals and be able to comply with traffic laws. 

Autonomous Trucks

The International Transport Forum (ITF) revealed in a report that autonomous trucks have a lot more to offer than traditional human-driven trucks regarding reduced cost, road safety, and lower emissions. In October 2010, Uber Advanced Technology Group (formerly Otto) completed the first autonomous truck delivery which delivered around 50,000 cans of beer over a distance of 120 miles.

Also, a Chinese startup, TuSimple, established in 2015 completed a 200-mile autonomous truck drive. The Chinese company claimed that its driving system was trained using deep learning to simulate miles of road driving.

Truck Platooning

This refers to the intelligent caravanning of two to five semi-trucks that can accelerate, steer and brake in a synchronized fashion. The intelligent caravanning is made possible with a machine to machine communication and cruise control technology, with a human driver operating the lead truck.  A platooning trial is set to take place in 2019 on the UK motorways sponsored by the British Transportation Research Library, DHL, and DAF trucks. 

4. Warehouse Management

Warehousing is an important part of any logistics company. To be successful in carrying out their business, they must have a good knowledge and understanding of the inventory, the demand, the supply as well as possible challenges. AI can provide intelligence from the data collected to provide vital pointers like flaws in supply, i.e., understocking or overstocking or susceptibility of some goods to damage when in storage.

In the logistics industry today, sorting of thousands and even millions of letters, parcels and shipment can be a challenging manual process that requires different components like an array of conveyors, infrastructure scanning, equipment handling and on hands personnel. 

The logistics industry can streamline this process by drawing on AI-driven robotics from the recycling industry. ZenRobotics, a Finnish company, developed a robotic waste sorting system that uses a combination of machine learning algorithms and computer vision to sort and pick recyclables from moving conveyor belts at a rate of 4000 items per hour 43. A similar solution for logistics can be applied to parcels and letter-sized shipments to reduce the effort and failure rate of human activity.

Wrapping Up

AI is not the future; it is now. With customers demand expectation being on the rise daily, the traditional method of process optimization and improvement might be unsuitable for addressing complex operational planning and structural sales issue. Therefore, it is more important than ever that companies deploy AI technologies for driving improvement and innovations in their transport and logistics operation.

Hope you like this article!

All the best,

Emmanuel

Please share your thoughts and views by leaving a comment. If you appreciate the insights of this article, please likeit and share it with your network. Many thanks!

Disclaimer: The views reflected in this article are the views of the authors and do not necessarily reflect the views of any company or organisation.

#AI #artificialintelligence #DataAnalytics #MachineLearning #DeepLearning #AdvancedAnalytics #Logistics #Transport