The Changing Dynamics of Logistics with Artificial Intelligence

Source: Artificial Intelligence on Medium

The Changing Dynamics of Logistics with Artificial Intelligence

We are living in a dynamic era; the digital revolution is redefining the way we live and the pivotal role in this revolution will be increasingly played by Artificial Intelligence. Artificial Intelligence can be defined as human intelligence being exhibited by machines; systems that approximate, mimic, replicate, automate and eventually improve on human thinking.

Machine learning is a subset of AI and deep learning is a deeper subset of AI.

Artificial Intelligence might one day function as well as human intelligence; machine learning is a specific term that refers to machines being designed to gather information (usually within a niche domain) and these machines learn on the basis of the inputs they are provided with.

Deep learning is a subset of machine learning. Hence, it is about learning continually; the system intends to learn from the observations based on real-world problems and adjusts the learning model as it takes in more information and new insights.


Logistics is increasingly being redefined by Tractica, an independent market research firm has estimated the sales of warehouse and logistics will reach $22.4B by 2021 and implementing AI across the supply chain can allow companies to garner $1.3trillion to $2trillion per year. AI is even helping companies save money and improvise business processes. Every year, United Parcel Service (UPS) is saving 10 million gallons of fuel due to AI.

So, how is AI shaping the logistics industry?

The business world is getting more complex and competitive with each passing day as companies operating global supply chains work under tremendous pressure to deliver better services at flat or even lower costs. In such scenarios, AI offers an opportunity to save time, reduce costs and increase productivity levels and accuracy with cognitive automation.

Source: Boston Consulting Group

DEMAND PREDICTION with Big Data: The beauty of AI is that the game-changing aspects of logistics; its ability to predict demand, predict trends, optimize delivery routes and manage networks. According to research by Deloitte, in many cases, algorithms done by AI can predict better outcomes than human experts. AI also aims to improve customer experience through

personalization, product suggestions, and customization tailored specifically to suit a user’s buying habits and preferences.

In February 2017, fidget spinners took over the social media with a blast. AI helped identify the quantitative rise in the talks around fidget spinners as well as the context of interest from a semantic understanding of unstructured talks on social media. AI along with Big Data enables predictions about fads which could boom into a fashion similar to fidget spinners.

AI out-performs expert human planners: In 2017, VersaFleet, a deep-tech operations research startup, pitted 3 companies’ veteran logistics planners of more than 10 years’ experience each, against its transport management system’s AI-based optimizer. This ‘human vs machine’ demonstration was inspired by IBM’s Deep Blue vs Gary Kasparov in chess (1997), and Google’s AlphaGo vs Lee Saedol in Go (2016). Just as in those landmark head-to-head match-ups, VersaFleet’s optimizer comprehensively outperformed all 3 companies’ human planners — it optimized in less than 1 minute what took each human expert more than 90 minutes to plan. In computer science terms, the vehicle routing problem in daily logistics planning is regarded as np-hard, meaning it is more difficult than chess and at least as difficult as Go. This comprehensive demonstration by VersaFleet proves that AI innovations have practical implementations in logistics, and has now become a norm.

Source: VersaFleet R&D labs, 2017
Source: VersaFleet R&D labs, 2017

DHL’s Global Trade Barometer is a unique and innovative tool for early indication of the current state and future development in global trade. The tool uses a large amount of operational logistics data, advanced statistical modeling, and artificial intelligence to project a monthly outlook for the global economy. The model works on a bottom-up approach and imports and exports data of intermediate and early-cycle supplies from seven countries that serve as the basis of input for the system, AI engine with other non-cognitive analytical models, expresses a single value to represent the weighted average of current trade growth and the upcoming two months of global trade. Tests with historical data reveal a high correlation between the DHL Global Trade Barometer and real containerized trade, providing an effective three-month outlook for global trade. (Source: IBM report: AI and Logistics)

DigitalGlobe, a satellite imagery company delivers high-resolution pictures of the planet’s surface to ride-sharing giant Uber. These images are rich input sources for the development of advanced mapping tools which help increase the precision of pick up, navigation, and drop off between the drivers and riders. DigitalGlobe’s satellites can find out new road-surface markings, lane information, and street-scale changes to traffic patterns before a city adds them to its official vector map.

DHL Parcel was one of the first delivery companies to offer a voice-based service to track parcels and provide shipment information with Amazon’s Alexa. Customers could simply ask Amazon Echo speaker “Alexa, where is my parcel?” or “Ask DHL, where is my parcel.” Customers can then mention their tracking number and receive shipment details. In case of any further details or queries, the customer could then call up the customer care for more information.

BACK OPERATIONS with Cognitive Automation: The combination of Robotic Process Automation (RPA) and Artificial Intelligence provides the employees/workers an opportunity to improve the quality of work and reduce the amount of time that is required to complete the allotted task. The daily chores are repetitive tasks, which can be easily automated. When the tasks are automated, the efficiency automatically increases; cost decreases, the accuracy of work increases and timeliness of data improves.

Cognitive Automation refers to business progression with a combination of AI and robotic process automation (RPA). Cognitive Automation replaces the work hitherto being by clerical labor using software bots that can be easily synced into existing business applications and IT systems.

RPA does not signify AI; AI learns and extracts insights and relevant data from unstructured data. Whereas, RPA executes rule-based workstreams, based on well-structured inputs on behalf of human workers, and has its limitation on learning beyond its initial programming.

Once the data is well-classified with the help of AI, an RPA bot takes it and uploads it into already existing accounting software to generate orders, execute payments, and send the customers a confirmation email, all without human intervention.

Source: Boston Consulting Group

UIPath has developed a robot that can impact approximately 99% of back-office operations since the robot can “see” screen elements.

LAST MILE OPTIMIZATION with AI and RPA: The ‘last mile’ in supply chains, the land-based distribution from distribution center to retail outlets or homes, is notoriously expensive and opaque. More than half of supply chain costs are sunk here (Honeywell, 2016) and it is common for parcels or even entire pallets of goods to be lost completely. The typical industry loss-rate for ‘proof of delivery’ (POD) is 20–30%, representing significant lost revenue for all logistics service providers (LSPs) involved. Modern transport management systems (TMS) like VersaFleet have been digitalizing these processes with electronic proofs-of-delivery (ePODs), removing the POD loss-rate to LSPs completely. Using a TMS like VersaFleet, Fortune 100 Brands like Canon, Johnson & Johnson and Philips have completely digitalized their last-mile logistics in Asia, accelerating their replenishment cycles and reducing stores’ stock-out occurrence to less than 50%. Built-in robotic process automation (RPA) instantly triggers SMS and email alerts based on pre-set rules, thereby eliminating the need for large customer service departments. With RPA and the ability to allow their customers to self-serve, each of these household brands’ customer service departments has since become twice more productive with half their previous headcount.

Source: VersaFleet, 2018

AUTOMATED VEHICLES with IoT: Driverless cars are no longer a dream, thanks to AI. The use of automated vehicles in the logistics industry guarantees to save time and money, and will almost definitely reduce the number of road accidents. Autonomous guided vehicles (AGVs) are already playing a role in logistics operations.

Fully automated Planning sans Humans: With a staff strength of 20,000 employees, Resorts World Sentosa (RWS), a Singapore-based integrated resort, previously required 61 buses to provide night transportation home for about 2,000 employees every night. Spending SGD 1.1M a year to provide this bus service, it used to hire planners just to manage the planning of ‘passenger to bus’ allocation for 4 waves of bus movements per night. Implementing the ‘People Transport Edition’ of VersaFleet’s optimizer, for more than 12 months continuously at the time of writing, zero human planners have been required — the AI-based engine has fully taken over the planning function entirely, freeing up human planners’ valuable time. The fully-automated optimizer not only plans in seconds what took each ‘human expert’ several hours but has reduced the number of buses RWS requires to less than 40, saving them hundreds of thousands of dollars in logistics costs. In fact, higher service levels are now achieved, with each employee being dropped off within 300m of their residence within 90 minutes, when before it was only within 500m within 110 minutes. In its corporate sustainability report, RWS cites this implementation specifically, quoting significant savings in energy consumption (in TJ) and carbon emissions (in kilotonnes CO2).

Source: VersaFleet, 2018
Source: Genting Singapore’s (Resorts World Sentosa) sustainability report

Waymo is working towards building driverless trucks, which will impact the logistics industry drastically.

Rolls-Royce, is working with Intel to develop self-driving cars. Driverless vehicles offer an opportunity to speed up the delivery process, optimize routes, reduce human error accidents, work 24/7, and more.

SMART ROADS with contextual intelligence: Another brilliant blend of logistics and AI is smart roads. Smart roads equipped with solar panel powered LED lights will alert the drivers about the conditions of the road. Smart roads equipped with fiber optic sensors can analyze traffic patterns. Smart Roads help make the roads traffic efficient, alert the drivers about road conditions and reduce the number of accidents.

WAREHOUSE AUTOMATION with Robotics: AI will impact many warehousing operations. Eg, data collection, inventory management, backend process and much more. AI can be used to predict the demand for particular products. Based on this data, modification of orders can be made on the fly and in-demand items can be easily replenished in the local warehouse.

Ocado, a supermarket in the United Kingdom has developed an automated warehouse. This sync of AI and warehouse automation is based on a robot called ‘hive-grid-machine’ which can execute 65,000 orders per week. The major tasks performed by the robot are moving, sorting and lifting products into the warehouse. Reducing labor cost and the time taken for order execution.

Human acceptance of AI in logistics: In the logistics scene, it is common to hear gripes from drivers ‘on the ground’, complaining about how the route plans from AI-based optimizers are “unrealistic” or not acceptable “in the real world”. To thoroughly investigate this issue, as part of a state-funded research & development program, VersaFleet worked with the School of Computing at the National University of Singapore (NUS) and ‘AI Singapore’, a statutory board that supports research & development in AI. Taking just one week of VersaFleet’s customers’ collective operations data, scrubbed and anonymized, it first tested whether the data-set was of acceptable size. Indeed, the entire map of Singapore could be mapped with this data-set. And with actually recorded driver speeds (tracked via each drivers’ smartphones) plotted by color intensity, it was, in fact, possible to visualize exactly where VersaFleet-enabled drivers slow down and stop to make deliveries (darker squares), and where Singapore’s highways are (lighter squares).

Source: VersaFleet, in collaboration with AI Singapore and NUS, 2018

To address the issue of whether human drivers are performing better with AI-based logistics optimization, this data-set was used to ask two important questions: a) How accurate is the estimated time of delivery (ETA) predictions by the optimizer?; and b) Do drivers fare better when they abide by system-suggested route plans?

For the former, actual date-time stamps of driver location history (logged by the minute from each of their smartphones) were compared to the AI-based optimizer’s ETA predictions. From the graph below, it’s visually distinct that the closer drivers abided to suggested route-plans (blue dots), the more accurate the ETA predictions, versus drivers who deviate from system-suggested routes (orange dots).

Source: VersaFleet in collaboration with AI Singapore and NUS, 2018

For the latter, similarly, actual date-time stamps of driver location history were used to assess how closely each driver manages to deliver within each delivery time window (tw_), e.g. if a time-definite delivery must be made within 9–11 am, what time did the driver actually arrive. The different colored dots represent drivers who abided to route plans (blue), versus drivers who significantly deviated from route plans (orange). From the graph below, it is clear that when drivers abided to system-suggested routes (bottom-left), they are more punctual and deliver on time (top-right), providing a superior logistics service.

Source: VersaFleet, in collaboration with AI Singapore and NUS, 2018.

Along with augmenting human capabilities, AI beholds an extremely promising future for the logistics sector and it might end up blessing humankind with unusual solutions that’ll revolutionize the technology in this era completely. AI will likely eliminate the routine mundane tasks.

Computer vision and language-focused AI will help logistics operators foresee, understand, and interact with the world in a more efficient and precise way. Similar AI technologies will help a new class of intelligent logistics assets that enhance human capabilities.

We believe the future of AI in logistics is filled with potential. As supply chain leaders continue their digital transformation journey, AI will become a core and inherent part of day-to-day business, accelerating the path towards a proactive, predictive, automated, and personalized future for logistics.