AI Archives

Cracking the Code for AI in the DC

Warehouse Management Systems | By Ken Ramoutar • 11/18/2021

There’s a new reason to optimize DC operations, and it’s bigger than the old reasons of productivity and efficiency gains. More and more companies are realizing that investing in their DCs and powering them with modern and sophisticated technologies like AI can lead to competitive advantages for the overall company.

This quest to master distribution has risen to the top of the list because customers (whether consumers or businesses) are demanding performance like never before. Expectations for short delivery times, 100% fulfillment accuracy, and large product selection and availability have become a key deciding factor on which customers make their purchase decisions. It’s apparent that Amazon has set the standard for consumer expectations, and those expectations have translated into the business-to-business world as well. On top of the insatiable desire for everyone to buy online fueling ecommerce growth in DCs, the high cost and scarcity of warehouse labor is crushing to many DC operators. The quest to deliver faster, better, and cheaper leaves operators turning to technologies like AI to try and crack the code.

AI holds the promise for big gains

The warehouse automation market is forecasted to grow at a CAGR of approximately 14% and be worth USD 30 billion by 2026. This broad category technology is driven by opportunities from autonomous vehicles, to robots, to automated storage systems, to software for optimization and orchestration of work.

Of the DC technology choices available, AI software is significantly promising for many looking to transform their DC operations. A recent study commissioned by Lucas Systems shows operators are optimistic about AI-based software, counting on quick and generous returns from their investment and expecting an average ROI of more than 60% within 5 years.

The fascination with AI technology and the large, expected payoff is driving a great deal of enthusiasm in many companies. Nearly 91% of survey respondents want to use advanced technologies such as AI and machine learning to drive warehouse and DC performance improvements. But they also admit that gaps exist in the knowledge and expertise required to achieve the full potential of AI. Approximately 80% of respondents agree that there is a general lack of understanding of how AI-based software can be put into practice to enhance warehouse/distribution center operations across multiple industries.

Getting started with AI in the DC

Getting going with AI without stumbling can be challenging. While leaders want the benefits of AI, they are nervous about having enough in-house knowledge and in some cases, giving up decision control to AI-based software. Here’s how to approach it:

Start with well-understood processes. You don’t need to know the nitty-gritty details of how AI and machine learning actually work behind the scenes. It’s more important to confidently understand your existing (non-AI) processes so that those processes can more easily be bolstered with AI.

Make sure your desired results are tangible and meaningful. Consider that projects which deliver hard labor savings or productivity or accuracy improvements often are perceived as more important than softer benefits such as employee satisfaction. What is your company’s top priority with AI?

Start with AI software that is well-proven out. AI development is still evolving but some areas have proven to deliver significant gains. Software for tasks such as slotting or labor management have AI engines that are well refined and require no AI expertise to use them.

Five fast start opportunities for AI

1. In-Warehouse Travel Optimization

The time spent navigating a warehouse can account for more than 30% of the total cost of labor. This makes the optimization of that travel time a critical component to running a successful warehouse or distribution center. Complicating the matter is that minimizing total travel time isn’t the only priority we need to take into consideration. The priority of each pick based on shipping deadlines can compete with travel minimization and also evolves in its level of importance throughout the day. For example, in the morning, we may want to maximize picking density to improve the efficiency of our pickers without much concern about shipping deadlines. Later in the day, those shipping deadlines become more important, so we need to adjust how we prioritize each pick, emphasizing the importance of shipping deadlines. Historically this problem has been solved using heuristic-based approaches, but it is a problem well suited for AI techniques.

2. Dynamic Slotting

Optimal placement of products within the warehouse, or slotting, has a significant impact on all the DC key performance indicators – productivity, shipping accuracy, inventory accuracy, warehouse order cycle time, and storage density. Yet, typical warehouses have fewer than a third of items located in optimal locations. This includes those that put effort into slotting via spreadsheet calculations and even slotting software packages. This is another problem well suited for AI with large expected returns.

3. Workflow Orchestration of People, Assets, and Machines

DC managers must consider how to optimize the assortment of systems and human labor and the orchestration of workflows between them. When you add robots to this mix, which systems will tell those robots where to go and when – and how to do it efficiently? Who will orchestrate the work between DC associates and the robots as work is handed off and another robot needs to step in? Who will reassign robots as the workload and order priorities change throughout the day? The key here is in analyzing data in real time – tasks which are well suited for AI.

4. Workforce Planning

As labor is typically the largest operational cost of a DC, workforce management plays a key role in cost-effective operations. Optimal allocation of workers to meet anticipated demand with each shift/day/week is essential to eliminate overstaffing and understaffing and reduce overtime while ensuring that orders get out on time. For example, during a shift, a supervisor may need to make a series of decisions related to shifting personnel from area to area to meet shipping deadlines associated with a wave of orders. Good decisions along these lines mean the supervisor will need a variety of near-real-time data in order to be alerted to and be able to process that data in a way that leads to good decisions. Again, a problem well suited for AI.

5. Performance Management

Engineered Labor Standards (ELS) and labor management systems have been around for some time, but an AI approach to performance management is better because it is easier to implement and produces better results. To implement ELS, industrial engineers typically go through a labor-intensive process that includes a combination of on-site observations, software calculations, benchmarking, and validations through actual practice. Additionally, ongoing re-evaluation is required in order to reflect changes in operations, such as when the type, weight, or size of products handled changes or when there are slotting changes. Finally, after all that, it is common practice to accept that the resulting standards are only accurate to within +/-15%, making it harder to hold individuals accountable. AI can make this a real time, dynamic process.

The next wave of DC transformation will be tech driven

Competitive advantage has shifted in many companies to be driven by the distribution arm of the company. And many companies have been static in evolving their distribution operations to meeting skyrocketing demands of consumers and businesses. Using sophisticated technologies like AI holds great promise for fueling the next wave of DC transformation. That tech is available today – so the key is to get going on the best opportunities.

—

Ken Ramoutar is CMO at Lucas Systems. In his 25 years of customer centric roles in supply chain software and consulting, Ken has navigated companies through uncertainty and volatility as a thought leader and change agent. Prior to Lucas, he was SVP and Global Head of Customer Experience for Avanade (a $3B Accenture and Microsoft owned company), and held leadership roles at IBM, Sterling Commerce, and SAP/Ariba.

Descartes Evolution 2019: Disruptive Technologies for the Customer Experience

Logistics Trends | By Chris Cunnane • 03/27/2019

I’m coming to you live from Naples, FL at the 2019 Descartes Evolution Conference. This was my third time attending the event and as always, it did not disappoint. This year marked the largest conference yet, with more than 650 attendees; in fact, it has outgrown its old location in West Palm Beach and from the looks of it, may need a bigger home next year. Yesterday’s sessions were about the big picture, while today’s sessions will allow me to listen to Descartes customers tell their stories. The big theme of yesterday was transforming the customer experience. I’ll have more at a later date on the customer case studies, but this article will focus on the big trends around supply chain management and their impact on the customer experience.

Descartes Global Logistics Network

During his keynote speech, Descartes CEO Ed Ryan discussed the foundation that is Descartes: the Global Logistics Network. The network connects carriers (ocean, air, ground, and rail transportation companies), shippers (retailers, manufacturers & distributors, and service providers), governments (customs agencies, regulatory bodies, and other agencies & associations), and intermediaries (customs brokers, freight brokers, forwarders/3PLs/NVOCCs, and payment agencies). Currently, Descartes has 20,000+ customers worldwide on the network, which processes 18.6+ billion transactions per year, and has grown 21 percent. On top of the network are Descartes’ value-added services: Routing, Mobile, & Telematics, Transportation Management, Customs and Regulatory Compliance, Broker & Forwarder Enterprise Systems, and Global Trade Content. These solutions all tie into the larger network.

Technology Trends

Ed Ryan also highlighted some of the emerging technology trends that are disrupting the supply chain industry. These are all trends that have been around in some form or another for a few years. Ryan pointed out that some of these applications are becoming mission critical for transforming both the supply chain and customer experience, while others are, in his opinion, not there yet.

For example, at the leading edge of transformation are the Internet of Things (IoT), Artificial Intelligence (AI), Cloud Computing, Information Security, and APIs. From a Descartes standpoint, IoT is a major driver. Ryan pointed out that IoT has become commonplace for most companies, especially at a base level (think Bluetooth headphones). From a TMS standpoint, IoT is all about tracking assets. Companies can use location data of their assets, whether that is a truck, ship, or even a specific container, to better serve their customers. A late or early truck means Descartes’ Route Planner can change the routes and inform customers of delays or a potential early arrival, which can help customers make more informed decisions.

AI is continuing to play a bigger role in helping customers make better and more targeted decisions. Two examples that Ryan gave were for the company’s routing solutions and MacroPoint. Within the routing solutions, AI has been used for years to determine how quickly a person can do their job. From a MacroPoint standpoint, AI is helping to give customers more accurate ETAs when a driver is still a few days away. They know how fast the driver drives, when he/she drives, and as he/she gets closer, the data will be better. For the customer, this is important as they need to know when the driver will arrive, so they know whether they need to keep a dock door open or schedule another truck to take that spot.

Cloud Computing, Information Security, and APIs are also becoming a bigger piece of the supply chain puzzle. As cloud computing is continuing to gain traction Descartes is moving more data to the cloud. While the company still operates their own data centers, more is being migrated to the cloud for scalability. In a session on technology, Ken Wood, Executive Vice President of Product Management at Descartes noted that cloud technology is becoming a fundamental technology enabler for the company. This adds a layer of security and accessibility.

Looking at the information security component, Ryan pointed out that companies today face hundreds or thousands of hacking attempts on a daily basis. As a result, Descartes has a full-time team dedicated to keeping its network safe from malicious activity. In the aforementioned technology session, Raimond Diederik, Executive Vice President of Product Management at Descartes noted that the full-time team is not an extension of IT. Instead, information security is in the forefront of their mind and is part of everything that Descartes does, including roadmaps, R&D, and corporate risk analysis.

APIs were the final technology trend Ryan pointed out. Before, when customers wanted to pull information and send it to their customers, it was done in the Descartes application which could cause integration issues. Now, the advancement of APIs allows customers to extract info and disseminate it to their customers in their own application.

Transforming the Customer Experience

I sat in on a session moderated by Chris JonesExecutive Vice President of Marketing & Services at Descartes, featuring Mark Parsons, Managing Director at BC Sands and Len James, CFO at John S. James Co. Chris Jones highlighted the 3 Rs of supply chain as developed by the late supply chain pioneer Donald Bowersox: Responsiveness, Reliability, and Relationship. Responsiveness is getting the customer; Reliability is keeping the customer; and Relationship is owning the customer. These all play a big part in the customer experience, especially as it relates to supply chain. Mark Parsons pointed out that retaining customers and getting new customers is the key to staying in business. To do this, you need that “WOW Factor.” And that comes down to not just delivering on your customer promise but going above and beyond. For BC Sands, this means sending notifications via SMS about the delivery status of an order. And if it is going to be late, getting ahead of the problem so customers can plan accordingly. As Parsons said, “the perception becomes that we have our s**t together, which keeps customers happy and coming back.”

Len James pointed out two major effects that have created ripples in the supply chain: the Amazon Effect and the Uber Effect. James said that “it used to be you can have things good, cheap, or fast; now you need to do all three. This is the Amazon Effect and it can take a toll on the business.” During a session on the vision for TMS, Brian Hodgson, Senior Vice President of Industry Solutions at Descartes expanded on the Amazon Effect. He asserted that Amazon is constantly innovating its delivery solutions. First there was free shipping, then Amazon Prime, then Prime Now, then big and bulky deliveries with set-up and installation, and finally, Prime Day, which allows customers to choose the day of the week they want all their packages bundled and delivered. This makes managing expectations and constantly delivering on-time more important. The Uber Effect is where everyone wants to know exactly where the truck is in real-time, and when it will arrive. Advanced visibility into trucks and containers becomes a core part of the supply chain.

Final Thoughts

Day one at Descartes Evolution did not disappoint. We have been covering disruptive technologies at Logistics Viewpoints, and it is always good to see the market take the same interest. Descartes CEO Ed Ryan pointed to a few significant technologies that are changing the supply chain landscape, notably IoT, AI, Cloud Computing, Information Security, and APIs. There are other trends that came up, but as Ryan indicated, he sees them as less impactful in the near-term, namely drones, autonomous trucks, and blockchain. However, he did go on to say that blockchain has its place and Descartes is running some pilots. But, from a digital ledger standpoint , especially for something like a bill of lading for example, it is simply too expensive and unnecessary from an encryption standpoint. And while autonomous trucks are still a ways off, the self-driving features will help trucks to be more safe in the next 10 years. From a technology standpoint, these trends are key drivers for transforming the customer experience. And as unlikely as it may sound, the supply chain may have the biggest impact on the overall customer experience. After all, if the customer doesn’t get what they want, when they want it, and how they want

A Trucking Firm Installs Cameras To Promote Safety And The Drivers Embrace The Technology? Preposterous!

Logistics Technologies | By Steve Banker • 10/18/2018

There is a driver shortage. Driver pay is increasing rapidly after years of stagnation. Drivers have choices. So, it came as a great surprise to me when Dale Willis, the Vice President of Data Services at Netradyne, said that his company’s camera and AI-based analytics system, used to improve driver safety, is appreciated by truck drivers! In my experience, drivers don’t like Big Brother. I was skeptical.

Media contacts for Netradyne did arrange a call with me with a driver, fleet safety executive, and safety training manager at Ward Transport and Logistics Corporation to attempt to confirm the outlandish allegation that drivers appreciated this technology. Ward Transport, a fourth-generation family owned company, primarily engages in less-than-truckload shipments. They are based in the middle Atlantic states. They have a fleet of 600 vehicles.

I first discussed the implementation of Netradyne’s Driveri platform with Steve Dunn who is the Director of Safety at the fleet. Ward Transport started looking at truck cameras several years ago. They had pilots with two different suppliers that did not work out. One solution’s cameras did not work well enough. In the other pilot, the supplier was able to show them after an event, for example a hard-braking incident, what they thought was an event. Ward Transport wanted a more proactive way to look at events. They wanted more control of what video they viewed and also wanted greater focus on positive behavior.

Enter Netradyne. When Mr. Dunn first saw the technology at a truck association conference, he was very impressed. But Netradyne was still a young company and he was not sure they could scale up and handle a fleet of their size. But after the second pilot failed, they went back to Netradyne and agreed to a 45-day pilot that began in January. That went well. In May they began installing the systems in 220 trucks across their three largest sites. By June that was completed. They now have three months of data, like what they have seen, and have preliminary plans to implement Driveri with the rest of their fleet next year.

The reason Ward Transport wants to expand to the rest of the fleet is that the system works well, it is the best video quality Mr. Dunn has seen. The camera, attached to the truck’s windshield, has cameras that face to the left and right, face forward, and face the driver. The cameras have a high enough resolution that in the event of an incident, reflected images picked up from the mirrors attached to the side of the truck can in many instances be viewed to determine what happened beside the vehicle.

Driver Safety Improves

It is increasing safe driving behaviors. Within two months after implementing the system stop sign events had decreased 60 percent. A moderate stop sign event is when a truck rolls through a stop sign without stopping at less than seven miles per hour. It is recorded as a major event if the driver goes through the intersection without stopping at seven miles per hour or more. The improvement in documentable safety has the executive staff interested. The next time they renegotiate their insurance they intend to use this data to see if they can get lower rates.

Finally, Ward Transport likes the fact that the feedback is not all negative, that positive events are reported, and drivers can be praised. For example, if a driver slows to let a car safely merge on the Interstate, this gets counted as a positive driving behavior. They believe having a system that includes positive reinforcement is key to driver acceptance.

Driver Safety — Dean White, a Driver at Ward Transport and Logistics Corporation

Drivers, Surprisingly, Accept the Technology

But do drivers accept it? Dean White, who is a driver with Ward Transport, says they do now. “Initially many drivers looked at this negatively. They feared that the company was spying on us.” Joe Drugo, the Manager of Safety Training, added “At first it was Big Brother watching. We were the Monsters on the Midway. I don’t hear that anymore. That is all gone now.”

One factor that led to increased acceptance was that the inward facing camera that could view the driver was disengaged. A cap was even put on this camera so that drivers knew that it could not secretly record them.

I asked, “on a five-point scale, with five being great, how would you rate it?” Mr. White said 4 out of 5. Mr. White has a positive view of the technology. “I’ve only been doing this for two years. There are many things I need to learn. I never turn down coaching. It makes me better. Having a camera in there, if I roll through a stop sign, I get reminded.” That helps to keep his focus on safety “every minute” he is on the road.

How did Mr. White think his fellow drivers would rate it? “For the fleet as a whole, the majority would rate it between a 2.5 and a 5. I was amazed that within a week of the staff recognizing positive performance on an internal TV in the break room, guys started taking pride in their scores.” The top scoring drivers are listed every week. “We began to have positive competition.”

Drivers can also see their scores and the events they are having on a driver app. When that app rolled out about two months ago, Mr. White said drivers were showing each other their scores and reviewing events from the day before. “This is real-time feedback. I like that.”

But a real change in driver acceptance occurred after an accident that occurred in Philadelphia. A car spun out in front of one of their trucks. The truck was able to stop, but there was still some contact. It was real good driving on the truck drivers’ part. But the car’s insurance company still tried to blame this on the driver and get the trucking firm to pay for the damage. When the insurance company was sent the video, that claim went away. That video was played on the television in the break room for the other drivers to view.

Dale Willis of Netradyne, by the way, took me on a test drive. To demonstrate the audio warnings the system generates, the ability to view video surrounding events for analysis afterwards, and the cool analytics, he engaged in some unsafe driving. This included zooming up to trucks on the interstate and closely tailgating them. Dale is very pleasant fellow, much more outgoing than I am, but I’m quite certain he enjoyed this outing far more than I did.

My impression from my drive along, and the technical conversation I had with Mr. Willis, is that what Netradyne has accomplished from a technological perspective is not easily replicable. A driver’s job is hard enough. The ability to incorporate more positive reinforcement into a safety program is a great idea.

Machine Learning in Today’s Warehouse Management Systems (WMS)

Cloud/Software-as-a-Service | By Clint Reiser • 04/04/2018

While speaking with a number of WMS product managers, I began to notice that machine learning was called out as a focal point for WMS product development efforts. In general, machine learning is a hot topic in the world of supply chain technologies. Just last week, Chris Cunnane wrote about machine learning for transportation execution. And Steve Banker recently wrote about Vecna Robots use of machine learning to improve its vision system. Naturally, I wanted to understand the relationship between machine learning and the traditional quantitative tools used in WMS. For example, how do they complement statistical methods such as regression analysis or optimization algorithms such as linear programming?

Machine Learning Qualities

Steve Banker published an article on Logistics Viewpoints that discussed three qualities that distinguish machine learning from predictive analytics . They are self-modification, high automation, and embedded within a process or workflow. These qualities provide a useful lens to view machine learning, its potential in WMS applications, and how it compares to other algorithms in use.

The quality of self-modification is what I consider to be a defining characteristic of machine learning for warehouse applications. It is especially useful for its ability to adapt to changing situations. Essentially, it is good for dynamic data sets, where the relationship between the dependent variable and the independent variables is fluid. When the predictive power of an algorithm changes, machine learning can recognize this degradation and create a new input-output relation that offers more robust predictive power. Markus Klug of SSI Schaefer explained the adaptive capabilities of machine learning to me by making an analogy to the more common approach of regression analysis, where a function tries to approximate the relationship between input and output variables by minimizing the error made. He said to imagine a multiple regression model that takes more than 100 variables into consideration, and also has the ability to take action if the explanatory power of the model (the R-squared) is dropping. In this analogy, the machine learning capabilities are designed to adapt the model to more accurately represent changing conditions in the warehouse by incorporating historical analogies or even altering the input variables taken into consideration.

Machine Learning Applied in WMS

Machine learning’s ability to adapt to changing conditions makes it especially well-aligned with the nature of today’s e-commerce warehouses that dynamically introduce orders into the existing work load. Waveless warehouse operations are a strong fit for the application of machine learning. Adam Kline of Manhattan Associates explained how Manhattan’s Order Streaming capabilities leverage machine learning to sense and adapt to changing conditions and priorities in the warehouse. Pick density may be the initial priority, but as carrier cut-off times approach, meeting order SLAs takes precedence. Machine learning is used to predict the time required to complete work. An optimization algorithm then uses those results to balance competing requirements while optimally utilizing available capacity.

HighJump is focusing on applying machine learning to waveless operations. According to Sean Elliott of HighJump, neural networks provide the ability to adapt to changing priorities as the operational environment changes. He believes the real value of machine learning is its automated ability to incorporate data points where a human wouldn’t adapt as quickly – essentially the ability to leverage data without reprogramming the system. I liked his way of describing how to identify opportunities to apply machine learning, “looking across the warehouse, where you find tribal knowledge is where machine learning shows promise.” The image of a neural network above illustrates how neural networks incorporate biases, or tribal knowledge, into their input-output relationships.

Oracle WMS Cloud is applying machine learning to dynamically adjust cloud computing capacity without downtime handle customers increasing volumes. Oracle WMS Cloud architecture and algorithms allow it to scale customer production server capacity as needed. For example, algorithms will evaluate the cloud computing needs for all customers across the globe and determine where additional computing capacity may be required by customers to support their business needs. Diego Pantoja-Navajas of Oracle explained how the complexity of dynamically managing server capacity for a large number of customers with diverse needs located across the globe lends itself well to the use of artificial intelligence (AI). In particular, AI assists in making sure that all appropriate variables are considered in the analysis. Oracle WMS Cloud applies the same type of AI to assist customers in the use of its WMS BI reporting tool. The tool analyzes data shared across the customer’s entire logistics network including inventory movements by location, user activity, vendor history, returns mix and disposition, lead times, and order performance. For example, the system may provide insights on trailer arrivals and the priority profiles of orders on each trailer and then prompt the customer, asking if they would like to prioritize certain trailers for expedited unloading at the dock. AI is helping Oracle WMS Cloud customers explore their data at deeper level and augment human analytical capabilities.

Marcus Klug of SSI Schaefer believes his firm is well-positioned to incorporate machine learning in its solutions because SSI Schaefer is responsible for highly automated warehouses where it controls the PLC to the material flow to the WMS. Among other possibilities, Marcus sees AI as well-suited to support the balanced movement of goods in the increasingly complex and dynamic warehouse, for example, by applying predictors to determine where congestion is likely to occur and subsequently avoiding it. But SSI Schaefer is taking a thorough and prudent approach to the introduction of artificial intelligence into its warehouse solutions due to the highly automated profile of warehouses in which it deploys its software and automation, and the critical role warehouses play in the supply chain.

Final Word

Although WMS suppliers are applying machine learning in their WMS solutions in diverse ways, they all agree that machine learning and AI offer vast potential to deliver additional value to their customers, and believe they are just beginning the journey into harvesting the value that can be delivered by intelligently incorporating AI in their WMS solutions.

Amazon Go – IoT to the 4th Degree

Inventory Management | By Clint Reiser • 01/31/2018

Amazon opened its check-out free Amazon Go convenience store to the general public last week. This high-tech low-touch store is a busy urban introvert’s dream. The opening triggered a second wave of media coverage that included descriptions of user experiences and a number of possible explanations about how Amazon is able to manage this high-tech customer experience. Amazon states on its website that this shopping experience is made possible by the same types of technologies used in self-driving cars. But I think it would be more accurate (but a less effective marketing message) to say that Amazon Go is enabled by the same technologies used in today’s high-tech warehouses.

How Does Amazon Go Do It?

Amazon isn’t providing many details about how it manages the checkout-free shopping process, but they do provide some insights into the technology elements at work, particularly the data capture tools. Amazon’s website states that Amazon Go utilizes computer vision technology and sensor fusion methods.

Customers must download a mobile application to their smartphones prior to becoming a store customer. This application facilitates payment and likely manages shopper identity and location while in the store. High-tech turnstiles manage the customer entry and exit processes. Sensors on product shelving monitor the movement of inventory as it is placed on and taken off the shelves. And perhaps most importantly, cameras capture images that can be used to monitor activity and make sense of data from other sensors used in the store. It is this pulling together, or “synchronization” of the data from various sources that appears to be the secret sauce in the Amazon Go recipe. I see similarities between the Amazon Go data synchronization process and PINC Solutions’ yard management system. PINC’s technology synchronizes RFID tag reads with location and time. A tag read alone simply confirms the presence of an inventory item. But the combination of tag read, location, and time allows the system to intelligently manage inventory, its location, and its change in status. I suspect that time stamps play a critical role in Amazon’s data synchronization process as well.

Computer Vision and Analytics at Work

When I first read articles about Amazon Go, I could envision how data such as inventory sensor, shopper/smartphone location, and change in inventory location or status could be synchronized to match product movement, and ultimately sales, with individual shoppers. However, this would require substantial assumptions (just because a person is next to a sandwich when it was picked doesn’t mean they bought it) and would likely provide insufficient transaction accuracy. After all, a few bad experiences could undermine the marketing power of the Amazon Go concept.

To obtain some expert opinions about how Amazon Go likely achieves accuracy and reliability, I reached out to Bernd Stoeger at Knapp, and Brendan McKenna at Zebra Technologies. Knapp developed KiSoft Vision, an image recognition technology that is applied to process improvement in distribution and manufacturing operations. So I knew he would have first-hand insights about the application of computer vision technology. He stated that the computer vision, sensor, and location technology may be deployed at Amazon go with an overlap in capabilities and serve to double-check the results of the multiple sensors. This overlap can serve to improve upon reliability and accuracy. For example, the image analysis may be able to confirm the simultaneous presence of a shopper and a product pick to validate the association between customer and item. However, this is not a trivial process to automate. There is a great deal of intelligence that must be embedded in the image processing technology to recognize and distinguish items and make associations in near real-time within a dynamic environment. The accurate and reliable association of an item to a shopper, and ultimately the purchase, is what Bernd views as a major complexity within the process.

Brendan McKenna at Zebra also provided me with his perspective on the likely technology in use at Amazon Go. Brendan drew upon his knowledge from the development of Zebra’s SmartLens technology. SmartLens uses multiple sensing technologies to make associations and ultimately deliver exceptional in-store inventory accuracy for retailers. The solution uses RFID, video, and micro-locationing. Zebra’s micro-locationing is a proprietary system that leverages ultrasonics. But Brendan from Zebra and Bernd from Knapp both stated that Amazon Go may leverage other micro-locationing sensors embedded in customers’ smartphones to obtain high levels of in store shopper location accuracy (better than GPS). Brendan informed me about the way in which multiple sensor technologies complement each other and improve upon reliability and accuracy. For example, video clearly requires line of sight, but Zebra’s ultrasonic micro-locationing doesn’t necessarily need line of sight to provide its data. He indicated that the analysis of the data streams and the correct association of those data streams require substantial programming logic and heuristics. At the same time, there is such a large magnitude of data that there is a need to filter out the unimportant to enable the analysis of that which is relevant.

I wish Amazon Go was located in Boston so I could easily visit the store and experience it for myself. Unfortunately, it is located in proximity to Amazon’s headquarters in Seattle, 3,000 miles from my home. But I haven’t given up hope yet. After all, Boston is still in the running for Amazon HQ2 – Amazon’s second headquarters in North America.

Artificial Intelligence in Demand Planning

Supply Chain Planning | By Steve Banker • 12/05/2017

While artificial intelligence (AI), particularly machine learning, has been used in supply chain applications for some time, there is an ongoing arms race to more effectively leverage both machine learning and artificial intelligence in demand planning solutions in new ways.

Demand Planning, a Key Supply Chain Planning Application

This is not surprising. Demand planning is one of the key applications in supply chain planning (SCP) suites. In ARC’s recent global market study on this market, demand applications account for just under a third of a $2 billion plus market. And these applications are often the wedge purchase; the SCP solution that is first implemented by a company that then goes on to purchase other solutions in the suite.

Artificial Intelligence in Demand Planning — *Machine Learning has been used to Improve Forecasting for Close to 20 Years*

Machine learning works by taking the output of an application (for example, a forecast), examining that output against some measure of the “truth,” and then adjusting the parameters or math involved in generating the output (forecast), and seeing if the adjustments lead to more accurate outputs. It can truly be said that the machine is learning from experience.

The definition of artificial intelligence (AI) is broader. Any device that can perceive its environment and takes actions that maximize its chance of success at some goal is engaged in some form of AI. AI includes expert systems that read and interpret written language or natural language processing applications like the iPhone’s Siri.

Using Artificial Intelligence to Clean Data

Before any demand management solution is implemented, the master data needs to be cleaned up. This process of cleaning and normalizing the data takes far longer, costs more, and requires more ongoing effort than almost all companies anticipate. AI is being used to help normalize the data. So, for example, demand planning applications can be used to forecast how much of a product will be shipped to a large customer. In the database, “Procter & Gamble,” “Porter & Gambel” (sp.), “P&G,” “Procter and Gamble,” and so forth all refer to the same customer. But digital applications need precision; All these names must be spelled and punctuated identically. It is an easy task, but a time consuming one, for a human to do. With AI, the task can be greatly sped up. LLamasoft is one solution provider that is using AI in this manner.

Cyrus Hadavi, the CEO at Adexa, said that data can be cleaned in another way. Forecasts often involve the inputs of sales people. “What we are doing is data testing. We look at past data and look for patterns. Which sales managers have been consistently too optimistic? Which one consistently too pessimistic? We build this analysis into the software.”

Artificial Intelligence in Demand Planning is Not New

JDA and Logility have been using machine learning for close to twenty years to help classify stock keeping units (SKUs) in demand planning applications. Different algorithms should be used for forecasting different types of products or groups. By properly classifying the SKU, better forecasts resulted. This has become common among supply chain planning suppliers.

Over time, more data inputs were introduced into the demand planning process, and many companies are doing far more forecasts over more planning horizons (daily, weekly, monthly, etc.) and ship to locations. E2open began to use this technology not to classify an SKU, but to find the downstream data sources that increase forecast accuracy for each ship to location and planning horizon. In short, E2open is using machine learning in more of an end-to-end black box type solution.

Other forecasts are possible. Sujit Singh, the Chief of Operations at Arkieva, told me that in the last three years customers have begun asking them to make sales predictions. The VP of Sales is saying, “For these key customers who have a sales order placement pattern, study their data to know when to expect the next order. Use this sales forecast to ask a customer that is 3 weeks late placing an order, “What’s up? Why haven’t you placed the order? Further, using a variation of market-basket analysis for the B2B world, sales teams are able to create demand for other products.” This can not only improve the demand forecast, it can help the sales organization close business.

Demand planning solutions improved when downstream data such as point of sale became available. But there are many, many more data streams that have not been tapped. ToolsGroup and Relex Solutions have done trials to examine whether using social media and weather data will improve forecasts. Eventually, these may turn into enhancements that move into their standard solutions.

A Need for Demand Modeling Tools

But many data sources will not be amenable to becoming part of a standard solution. A large paper distributor, for example, might be able to improve their forecast by using paper exports from Russia, but almost no other company could productively use that data. LLamasoft has introduced a demand modeling tool called Demand Guru to help companies develop more customized demand planning solutions. This tool allows a company to more easily clean, blend and then ingest time series data and then test whether that data set improves forecasting over some forecast period or ship to location. Historical third-party data can be imported into the tool and then the company can examine whether their past forecasts could have been improved by using that data set. The data could be anything from GDP data, demographic data, gas prices, housing starts, to the afore mentioned paper exports from Russia. It really is any time series data series that a forecaster believes might improve their forecast accuracy.

A few supply chain planning software companies are exploring the use of natural language processing to improve the user experience. Karin Bursa, an executive vice president at Logility, foresees a day in the near future when a demand planner asks the application at the beginning of the day, “What are the top three issues I need to address today?” Or, perhaps, “what are the top three promotions I should run to address declining sales in the southeast?”

A complete analysis of logistics trends, technologies, and services