Gain Visibility Into Every Corner of Warehouse Operations
February 18, 2021
Applying Machine Learning and Artificial Intelligence to Fulfillment Systems
The array of material handling equipment stacked, racked and running inside e-commerce distribution centers (DCs) is staggering. Automated, high-speed storage and retrieval systems. Robotics. Manual workstations. They all serve specific purposes within the fulfillment process, yet are interdependent; a disruption or malfunction in any one of these critical components can cascade throughout an operation.
Every machine, process and worker must be precisely orchestrated together in one harmonious symphony of activities. Otherwise, efficiency and productivity decline — and costs escalate.
The complexity of material handling interactions and their impacts on the entire fulfillment ecosystem are often underestimated and misunderstood. To achieve throughput targets consistently, meet customer service level agreements (SLAs), and maximize profitability, DC operators need advanced warehouse automation software that provides visibility into each corner of the warehouse.
That’s the role for next-generation warehouse execution systems (WES) with advanced data-driven optimization and machine-learning algorithms that provide real-time, automated decision-making — or what we call decision intelligence.
Accurately Predict Processing and Order Sequencing
Throughout a DC, powerful software platforms can collect and analyze data from every connected system and process. With the information already available, decision intelligence enables DC operators to improve performance within the following critical order fulfillment functions:
- Just-in-time, goods-to-operator (GTO) put wall and allocation — When an influx of orders consists of multiple items, both automated and traditional picking processes must be sequenced properly to coordinate efficient order consolidation precisely — either at a GTO station or a put wall. With ever-changing order fulfillment priorities and dynamic workloads, it’s no longer efficient simply to pre-assign orders and/or SKUs to designated GTO stations and put walls.
- Labor management — A WES with machine learning capabilities can evaluate individual picking rates and tasks in the queue to forecast and predict labor requirements and optimize fulfillment workflows accordingly. Armed with this intelligence, DC operators can allocate the optimum number of resources to workstations and pick zones based on anticipated or real-time workloads.
- Order prioritization and release — Many e-retailers balance delivery windows and customer SLAs continually, as well as fulfill a combination of in-store and direct-to-consumer orders. A modern WES weighs every factor to determine the most optimal release timing, fulfillment process, picking order and execution path. With decision intelligence, WES software can predict order processing time accurately, track system capacity, and provide optimal release sequencing to ensure on-time shipment and maximum profitability. Decision intelligence makes it possible to balance the workload continuously throughout the entire fulfillment system.
- Routing — Carton, tote and item routing based on a license plate number (LPN) are foundational DC functionalities, often within the capabilities of warehouse control systems (WCS). A sophisticated WES with decision intelligence impacts larger business flows by tracking the contents within each tote — not just its LPN — to inform real-time, dynamic routing decisions based on the next-best destination. At every routing decision point, these intelligent software platforms can act within milliseconds to route SKUs or orders to improve the accuracy and productivity of the operation and adapt to dynamic order priorities.
- Storage allocation — In dynamic automated storage and retrieval systems (AS/RS), machine-learning algorithms provide the decision intelligence to maximize storage utilization and enable smart slotting while ensuring optimal retrieval time frames per item. This advanced logic can determine optimal storage location dynamically for items of various sizes while weighing the trade-off between space utilization and retrieval times based on an item’s relative demand.
In e-commerce fulfillment environments where order priorities are constantly in flux, the machine-learning technology underpinning decision intelligence enables flexible, dynamic order reprioritization. For example, a WES quickly can locate an item that has already been picked for a lower-priority order, re-route it from the sortation system, and re-assign it to a higher-priority order — without impacting the shipment time of the original order.
Machine learning also provides the decision intelligence to monitor workloads at all GTO stations continuously and predict the times it will take for individual items to be picked for multiple orders — often sourced from a variety of pick processes throughout the DC. The software utilizes this information to coordinate the pick timing of items dynamically with their respective orders, and then allocate them to the most optimal GTO stations.
Similarly, for put wall order consolidation, machine-learning algorithms evaluate each picking task needed for active orders and then calculate the time it will take for items to arrive at the available put wall locations. By evaluating multiple fulfillment scenarios, decision intelligence generates the best picking and order release sequences to coordinate the arrival of orders and their items at the most optimal put wall stations.
Decision Intelligence Supports Real-Time DC Optimization
Matching the pace and complexity of e-commerce fulfillment requires new approaches that enable real-time, decision-making precision and much tighter control over dynamic DC operations.
The Momentum software suite from Honeywell Intelligrated is equipped with a powerful machine-learning engine — Decision Intelligence — that delivers the efficiencies required to optimize today’s e-commerce distribution centers. Equipped with robust WES capabilities, modular functionality and a powerful Decision Intelligence machine-learning engine, we’re helping leading e-retailers accelerate their digital transformations and empower fulfillment performance execution.
Our agile, extensible and state-of-the art software architecture is designed to allow you to scale up quickly to meet your current and future requirements. With Momentum software, you can combine advanced automation, robotics, AS/RS, voice, labor management, machine control, real-time asset monitoring and a wide variety of order fulfillment technologies — all within a connected infrastructure seamlessly. As a result, DC operators gain visibility and can take control over every facet of fulfillment operations.
For more information, read our white paper, Leveraging Data Science for “Decision Intelligence” in the DC.
To subscribe to our blog and receive exclusive communications and updates from Honeywell Intelligrated, click here.