AI in Supply Chain and Logistics
Introduction to supply chain
Recently, managing supply chains has grown much more difficult due to increased physical flows reflecting the increasingly complex product portfolio. The pandemic’s impact on market volatility has raised the need for adaptability and agility. Organizations and shareholders are putting more emphasis on supply chain resilience as the environmental impact of supply chains becomes more widely recognized.
Artificial intelligence (AI)-based supply chain solutions are anticipated to be a potent tool to help businesses overcome challenges. Artificial intelligence has changed the game in a number of different industrial verticals thanks to its ability to analyze huge amounts of data, comprehend relationships, provide operation visibility, and promote smarter decision-making.
A few of the main challenges that AI solves are:
- Demand forecasting
- Product scheduling
- Warehouse automation and management
- Segmentation or clustering of products
- Shipment load monitoring
- Defect inspection and quality control
- Intrusion and theft detection
Let's try to understand the challenges in depth and how AI solves these challenges
1. Demand Forecasting
Below is the difference between the traditional forecasting methods and machine learning forecasting methods
Table 1: Comparison of Traditional and Machine Learning Forecasting solutions
It is easy to identify that machine learning way is preferable and easier. Companies that used this saw the following improvements
- Errors in supply chain networks can be reduced by 30 to 50% with AI-powered demand forecasting.
- Warehousing costs decrease by around 10 to 40%.
- The loss in sales due to inventory out-of-stock situations can be reduced by up to 65% with improved accuracy.
- In general, the estimated impact of AI is between 1.2 and 2 trillion dollars in the manufacturing and supply chain planning
For Danone Group, AI in demand planning ultimately led to a 30% reduction in lost sales, a 30% reduction in product obsolescence, a 20% reduction in wrong forecasts, and a 50% reduction in the workload of demand planners.
2. Product Scheduling
It is very easy to organize the various requirements and deadlines using traditional planning techniques, such as material requirements planning and its derivatives. They create a plan or schedule that seems logical and might actually function by applying a fixed logic to the vast amount of data. A typical computer-generated plan, however, may include elements that are unrealistic or outright impossible, such as having a work center commit to three or four times the amount of production that it can complete in a day. This is something that any experienced planner will tell you.
The latest generation of so-called advanced planning and scheduling (APS) software, which trades off priorities and alternatives instead of using fixed logic to build a more workable plan that balances resources and material dates and amounts, overcomes many of these restrictions. Even APS, despite how “smart” it is, has its limitations. To evaluate and apply the recommendations in the context of what actually occurs on the shop floor and in the world of the consumer, utilizing insight that
The use of the word intelligence in the preceding paragraph is significant because AI is designed to do human-like thinking. In order to create a process model, these production planning systems “learn” by gathering enormous volumes of data and carefully analyzing it for cause and effect. The system may then run many what-if scenarios to determine the optimal course of action when new information becomes available. The model is improved in response to new information as these possibilities come to pass. Plans and timetables generated by AI will surely be “better” than anything a human could produce. Plants will function more effectively, products will be of higher quality, and more work will be finished on schedule and with less expense.
3. Warehouse automation and management
Warehouse management automation allows you to reduce (or eliminate) labor-intensive tasks like data entry and picking, reducing injury and fatigue while increasing productivity — and saving you money.
According to a 2018 report from the US Bureau of Labor Statistics, labor expenses can cost a warehouse up to $3,700,000 per year. Warehouse automation can save you money, which is especially important at a time when supply chain issues are on the rise, leading to an increase in operational costs.
Inventory management and data collection are two aspects of warehouse automation. And, while the initial investment in AI technologies for the supply chain will be costly, the long-term benefits will far outweigh any short-term losses.
Let’s take a look at some essential AI use cases in transforming warehousing processes:
- Goods-to-Person Systems (GTP)
GTP is a type of automated storage and retrieval system (AS/RS) in which items are delivered to or retrieved from specific storage locations by automated vehicles known as shuttles. GTP is a popular robotic process automation method for reducing congestion while increasing efficiency. Vertical lift systems and conveyors, for example, can more than double the speed of your warehouse picking when properly implemented.
2. Automated Guided Vehicles
AGVs (Automated Guided Vehicles) are a type of self-guided mechanized automation vehicle that uses very little onboard computing power. They use sensors (or wires or strips in some cases) to help them dash around the warehouse on a fixed path, transporting materials and delivering goods. However, using AGVs makes the most sense if you have a large warehouse with plenty of space. AGVs are less useful if your warehouse is small and there is a lot of human traffic.
3. Voice Picking and Tasking
Pick-by-voice (also known as voice-directed warehouse procedures) is a method of determining the best pick paths on a warehouse floor by combining mobile headsets with speech recognition software.
The system then relays all necessary information to the warehouse worker (such as the quantity required for an item) before guiding them down the correct path so that they can either pick out a product or store one. The method eliminates the need for a handheld device (such as RF scanner), allowing a picker to concentrate on the task at hand. It also increases efficiency, safety, and order accuracy.
4. Segmentation or clustering of products
Algorithms like K-means clustering and density-based clustering are widely used in industries to reduce manual labor expenses. Using barcode scanners, sensors, and RFID, Automated Sortation Systems locate products on a conveyor before moving them to a certain warehouse location. An automated sortation system can be used by businesses to enhance order fulfillment while doing away with the requirement for human monitoring. Improvements may be made to receiving, packing, and shipping, and Amazon is one of several major businesses that has already deployed robots to pick and sort.
Automated sortation systems are classified into two types:
- Sorters for units
- Sorters of cases
For both categories, Multiple sortation systems drop items onto a conveyor belt before diverting them to the appropriate location in the warehouse. AI-based systems reduce your reliance on labor, improve order accuracy, and increase productivity and efficiency
5. Shipment load monitoring
By late 2022, COVID-19’s harshest effects will be behind us. The shipping sector, particularly ports, has been hit so hard by the pandemic that recovery is taking a very long period. Supply chains stall when the shipping industry is knocked off balance.
AI-based logistics solutions are the perfect option as the shipping sector is rebuilt. Monitoring of container loading is a key priority. It is a company that manages the process of packing goods into shipping containers. Before the container is sealed and transported to the port, this is typically done at the same manufacturing facility where the goods were created. After that, it is exported.
How can artificial intelligence help?
The importer can have 100% assurance from artificial intelligence that they are getting exactly what they intended. AI systems, for instance, can carefully watch the loading of freight, seeking fragile or highly expensive things.
Pre-shipment inspections on the importer’s end are possible (and do happen), but they rarely guarantee that the items loaded onto the containers are the ones that have been inspected. For instance, the supplier might swap the products after an inspection. Implementing container loading monitoring helps hold your suppliers accountable if there is a lack of confidence between you and them.
You can quickly and accurately identify defects discovered during container load monitoring thanks to the use of AI in the logistics industry. The software can detect the following common flaws:
- Rejected carton boxes
- Broken wooden pallets
- Incorrect pallet stacking
- Non-inspected lots
6. Defect inspection and quality control
When it comes to the quality of the products they buy, consumers are more demanding than ever. As a result, defect inspection and quality control are critical processes in the logistics industry.
Although humans usually execute this activity, its forensic nature makes it time-consuming and susceptible to human error, primarily as a result of mental weariness brought on by having to perform repetitive tasks. When mistakes are made by humans, customers are unsatisfied and flaws are undetected.
AI in logistics can help with quality control and defect inspection. It not only speeds up the procedure but also makes it more accurate. AI-based techniques can combine digital image processing, picture classification, image segmentation, and computer vision to find flaws that the human eye would first overlook. As a result, the inspection process is more efficient, and the number of defective products sent to consumers is greatly reduced.
There are a number of problems here that AI in logistics can overcome:
- Incomplete or Delayed Shipments
Cameras can detect whether or not shipping labels are present on packages. If the labels are illegible, they can also send notifications. When AI software can automatically scan labels during the loading process, the need for human labor is eliminated.
2. Damaged Items
Numerous businesses have produced AI models that employ object detection to identify faulty products on the production line. These businesses considerably cut the number of hours worked thanks to their AI models. Fujitsu is one of these businesses.
Because of how sophisticated Fujitsu’s model is, the Japanese business was able to create training data without using actual photographs of damaged products. Instead, they used pictures of objects with fictitious faults to train their model. However, their model is still able to comprehend all of the characteristics of the objects it will be examining. AI systems can also recognize when a package is dropped while being transported (and therefore potentially damaged).
3. Poor Packaging
Poor packaging might also result in damaged goods. Your products are less likely to be harmed during supply chain operations if they are appropriately wrapped. Your products will be handled attentively if they are packaged well, especially during loading and unloading. As a result, there are fewer returns and a better customer experience.
You may optimize your product packaging with the aid of AI algorithms. AI may help logistics businesses create better packages for certain products, choose the right materials and sizes for each delivery, and notify you before a box is dispatched if it has been tampered with. Preventing delays along the way effectively lowers the likelihood of supply chain issues.
7. Intrusion and theft detection
It is critical that you consider security measures that prevent intrusion and theft, thereby saving you money. AI is crucial in detecting intrusion and theft. The technology can monitor security camera activity, looking for anything suspicious and identifying objects of interest.
Artificial intelligence-powered intrusion detection systems recognize objects based on their location, size, and movement. They go beyond standard intrusion detection systems by utilizing a more sophisticated algorithm to recognize various object types while reducing the number of false positives.
Some practical applications:
- Automatic License Plate Recognition
This is a smart infrastructure computer vision system that alerts you to any vehicles on your property that aren’t supposed to be there. Automatic License Plate Recognition (ALPR) allows cameras to read license plate numbers quickly and with no human intervention.
ALPR is used by security and government agencies to prevent crime and can be used in conjunction with object tracking and facial recognition to deter criminal activity. Automatic license plate recognition is also used in electronic toll collection systems to charge vehicles entering and exiting roads.
2. Video Analytics
Video analytics is a technology that processes a digital video signal using a special algorithm to perform a security-related function. There are three common types of video analytics:
- Fixed algorithm analytics
- Artificial intelligence learning algorithms
- Facial recognition systems
Video analytics allows you to detect and validate issues before taking action. It essentially provides you with actionable data on activity around your facility, allowing you to quickly identify potential intruders. Importantly, video analytics can identify the movements of vehicles and people in a scene while ignoring irrelevant motion. It can aid in the detection of unusual motion and unusual activity.
Access control systems that work in tandem with intrusion detection sensors to generate real-time alerts whenever an unauthorized person attempts to enter your facility are important areas of video analytics. This is extremely helpful in reducing intrusions and theft. It detects faces using computer vision, with facial data stored in a database and is one of the most widely used AI applications in logistics.
Meanwhile, intrusion detection sensors detect movement before sounding an alarm if an intruder is detected. Data is collected and annotated, neural networks are trained, and deep learning algorithms easily (and automatically) highlight objects of interest. Objects of interest are typically identified within seconds of a person entering a specific area.
Cons of Using AI in supply chain
- The high cost of AI implementation in supply chain management.
- The lack of transparency and control over the AI-powered supply chain.
- The need for specialized skills and knowledge to operate AI in supply chain management.
- The potential for disruptions in the supply chain due to AI.
- The impact of AI on jobs in the supply chain management industry.
CONCLUSION
Artificial Intelligence continues the journey of developing digitalization and becoming a bigger and more essential part of daily business.
In industries like logistics, AI that receives from experience is a helpful method for determining significant problems. Artificial Intelligence is operating a significant role in stimulating the path toward a predictive, proactive, and personalized prospect for logistics.
Authors: Rishab Khandelwal, Rajat Harne, Saharsh Samir, Tanmay Pol