Specialists in logistics and supply chain management face multiple sets of contradictory requirements: Reduce the space they use, but increase the volume of goods they distribute; decrease latency while lowering transport and handling costs; and, most conspicuously, cut labor costs but maintain or increase service levels.
Unlike corporate functions such as manufacturing, distribution cannot be entirely outsourced to countries that have low labor and real estate costs. Some hubs and packaging centers can be located in lowcost areas, but distribution centers (DCs) need to be situated in or near the areas they serve. Labor costs are a reflection of the labor market in those specific locations.
What’s a DC manager to do? Automation is both the key and the opportunity for companies that can deliver solutions to increase productivity and reduce costs while allowing for maximum system flexibility.
Of the range of technologies required to achieve total automation, this article focuses on two approaches to robotic automation that could have the greatest impact on warehouse management and supply chain efficiency: mobile robotic warehouse systems and automated guided vehicles (AGVs).
Left with few options to cut costs and increase productivity, warehouse and distribution managers have embraced robotic automation in far greater numbers than other industries, with the exception of heavy manufacturing. In addition, the very broad range of warehouse automation tasks demands that managers adopt systems that are far more advanced, with respect to learning new tasks, identifying targets, and combining subroutines, than the typical manufacturing system.
During the past two decades corporate logistics and supply chain groups have adopted a broad range of technologies, far sooner and often to far greater effect than other corporate departments. GPs systems first became widely used in civilian applications when trucking and shipping companies began fixing satellite- or cell-phone-enabled units to trucks and containers so that locations could be reported in real time. RFID, now included in many credit and debit cards, retail packaging, and pet-identification implants, first became successful as a way to tag and track shipments, providing warehouse managers with a simpler way to locate specific loads within warehouses and log them in and out of the facility automatically. Distribution centers were also among the first to adopt handheld computer units to automate picking and record keeping, computer-controlled conveyor systems using barcodes to identify packages, and robotic systems capable of packaging individual products.
Automation has become so embedded within warehouses and DCs that it is unusual to find operations of any size that rely solely on human labor. But the physical movement of goods and materials is only part of the overall automation picture. EntERPrise resource planning (ERP) software such as SAP’s R3, which provides modules for orders, billing, location, and other logistical functions, is also a key part of any warehouse management system (WMS).
It is common to find specialized automation technology integrated into systems automating these other business processes. Warehouse automation often involves the integration of ERP billing, shipping, and scheduling information to such a degree that it resembles ERP functions more than robotics systems.
Room for Improvement
The automation picture within the physical side of the operation—the actual moving of goods—is far less integrated and less settled. ERP or WMS systems deliver orders to the distribution center in formats and sequences that make the work more efficient. However, specific orders are often delayed because the pickers, or transport they are assigned to, are backed up with other work.
Automated storage/retrieval systems can considerably reduce the time it takes to assemble an order. For example, moving products from upper shelves to more convenient lower ones can decrease the time to fill an order by one-half, while yard management systems can ease transport congestion outside the DC. Automated inventory systems speed the product check-in process as well as increase accuracy.
Most of the pick-pack-and-ship operations are still conducted by humans. They are often helped by pick-to-light or other assistive systems that make their work more efficient, but without the full automation and integration that could deliver efficient, low-labor-cost, scalable DC automation systems.
even at the current state of the art—systems assisting humans, rather than humans directing systems—best-in-class distribution centers are able to achieve 95 percent on-time deliveries one and a half times more often than average, according to a survey of 250 supply chain executives published by Aberdeen Group. Automated and efficient warehouses in the survey were 76 percent more likely to boost inventory accuracy to 99 percent or higher, 36 percent more likely to have reduced labor costs an average of 3 percent per year, and 40 percent more likely to consistently ship within one day of an order’s placement.
Full automation would improve those figures—especially the cost savings—considerably. It would also enable automated DCs to support increasing volumes of international shipments and direct-to-home shipments that continue to grow along with orders over the internet. By knowing the location and condition of every item in their supply chains, fully automated systems would also be able to quickly adapt to ever more complex shipping and security regulations.
The ultimate goal is a fully automated warehouse—a system that could receive shipments, receive orders, and fulfill them with little or no intervention from humans. There are many elements to such a system, including:
- Automated Storage/Retrieval Systems (ASRS). Cranes or lifters that move goods from one shelf to another to maximize use of space and increase the convenience of retrieval.
- Carousels. A combination of conveyors and AsRs that allows goods to remain on a single shelf, while the shelf itself revolves into the correct position for sorting.
- Conveyors. Conveyor belts, boxes, or cars on rails that deliver goods from one section of the DC to another.
- Sorters. Identify products and mark them for placement or place them in conveyors to be moved.
- RFID, Bar Codes. Radio-enabled or visual coding that allows other systems to identify an item.
- Pick/Pack/Palletize. Robots, usually in fixed positions, that assemble packages of large items, assemble whole pallets for delivery to a single location, or move large numbers of items from storage to a staging or picking area.
- Veriflcation Systems. Barcode or RFID readers, scales, or other mechanisms to identify and verify a package is where and what it is supposed to be. Inconsistencies in weight or label identification can cause items to be rejected.
- Scalable Case Picking. Planning systems that analyze order data and assemble pallets optimized for the most efficient picking and shipment to specific customers.
- Software. Order entry, billing, operations, and integration of data on orders, items, destinations, and packaging that provides orders to the DC systems.
- Integration. Not all commercial automated systems and software can exchange data effectively, presenting gaps in end-to-end automation.
Robots and AGVs
Automated guided vehicles or automatically guided vehicles (AGVs) are the most common form of robotic assistance in warehouse and DC automation systems. They are also common to manufacturing environments. AGVs range from enormous factory units designed to move multi-ton products from one stage of fabrication to another, to chair-sized units that navigate hallways carrying smaller amount of materials.
Recently, AGVs in warehouses and distribution centers have been joined by a new class of intelligent, mobile robots (sometimes referred to as intelligent AGVs). These systems are typified by new methods of navigation and increasing levels of autonomous behavior.
Regardless of the form, it is well accepted that robotic automation for manufacturing, warehouse, and distribution operations has many benefits. For example, in a recent study by RMT Robotics and Modern Materials Handling magazine on the role of AGVs, materials handling professionals were clear in their belief that AGVs increase productivity (Figure 1) and can be applied widely (Figure 2).
Guidance and Navigation
AGVs perform a range of functions, but often in a rigidly prescribed way. Using vision, magnetic detection, wire, or laser-identification systems, they move from one place to another—usually within a single building and on a single floor, though they can be programmed to ride elevators, take ramps, or use other mechanisms to reach other floors. AGVs are easy to reprogram, and are widely used in warehouses and DCs. Rather than require several pickers to move the packages they have assembled to a distant part of the DC, AGVs can be used like a shuttle: Human pickers load packages on the AGV, hit the Go button, and the AGV goes wherever its guideline takes it.
One major advantage of AGVs is the ease with which their destination programming can be changed. Change the line and you change the end point of an AGV’s trip. That is a critical feature in DCs that shift the location of products, must accommodate frequent changes in stockkeeping units (sKus) or the size of packages, or support a range of loading bays as a destination.
More advanced navigation systems do away with tape on the floor in favor of location awareness of various types. The most sophisticated is referred to as natural-features navigation, in which the “natural features” are most often radio or visual beacons placed at a height so that the AGV can “see” enough of them at any time to be able to triangulate its own position. Natural-features guidance systems can absorb new routes and new commands more easily than systems that rely on markings on the floor that have to be altered in order to provide new routes.
AGVs use laser-detection systems or inertial guidance systems to identify their progress and location. In some cases they might employ GPs receivers, though reception problems within a DC and precision rates of three meters or more make inside use of GPs-based systems problematic.
Other guidance systems include gyroscopic inertial guidance, a highly accurate “dead reckoning” navigation system in which the AGV measures the distance and direction it has gone to determine its location. Shortrange transponders on DC walls and location-identifying receivers on AGVs are also used, though the margins for error are greater than for either inertial guidance or lasers.
AGVs have a long history in warehouses and distribution centers, beginning in the 1950s and continuing today. During that time period the number of companies offering AGV systems increased, as did the number of installations. The systems also increased in sophistication. For example, navigation systems evolved from simple wire following to laser-based (the majority today).
System flexibility also advanced over the years, but it was largely achieved through modularity. Today, AGVs from established companies such as RMT Robotics Ltd., Jervis B. Webb Co., Egemin Automation inc., and others come in a wide variety of sizes and configurations, from retrofitted “smart” forklifts or jacks, to the robotic equivalent of a flatbed—a large guidance and motor unit in front with an open, flat freight area behind. It is quite common for AGV providers to offer a few standard platforms that can be easily customized for specific materials handling operations.
AGVs designed to directly carry products are most common, but tug-type units are growing in popularity because of their increased flexibility. Rather than having a platform on which goods can be piled, or forks that can lift a pallet or other platform, tug AGVs tow other vehicles. That enables DC managers to use a variety of heavy- and light-duty tow vehicles, sometimes allowing for the use of a smaller number of AGVs by closely matching load requirements to the available AGV platforms. The new generation of intelligent, mobile robotics systems for warehouse automation that have recently entered the market are physically modularized to a much lesser degree, but through a combination of intelligent software and advanced navigational techniques they have proved to be flexible automation solutions. For their part, too, the established companies continue to innovate, often by extending the “automation chain” as they work toward full automation.
The following are three snapshots of both emergent and established providers of mobile, warehouse automation technology. Each example serves to illustrate how companies are offering solutions that increase system flexibility, integrate with other automation technology, and extend the warehouse/DC automation chain.
Jervis B. Webb: SmartLoader
One of the current leaders among established AGV providers is Farmington Hills, Mich.-based Jervis B. Webb. The firm’s SmartLoader solution is designed to move loads from freight delivery or on-floor storage straight into trucks for delivery. Laser radar (ladar) guidance positions the SmartLoader inside the truck to minimize the requirement for human intervention at the last moment.
Webb offers a range of modular towing vehicles capable of pulling loads of 50,000 pounds and that can be customized with conveyors, tilt beds, and other load-enabling equipment. Its units are guided by transponders buried in the floor, and a management application gives DC managers control of several vehicles at once. More importantly, the company provides software hooks that allow its guidance systems to be integrated with warehouse management software, providing much greater potential for end-to-end auto-mation than offering software to manage the AGVs alone.
Kiva Systems: Mobile Fulflllment System
while AGVs provide an immediate improvement in efficiency by cutting travel time, systems with more intelligence and greater adaptability promise to deliver efficiency in other aspects of DC management as well. For example, woburn, Mass.-based Kiva Systems Inc. combines transport and storage with a robotic unit that slides under entire shelves of products and carries them to the shipping point. High-tech, high-volume DCs with very short ship-time goals such as Zappos.com, The Gap, and Staples all use Kiva picking and delivery systems.
Kiva was founded in 2003 by Mick Mountz, an MiT-trained engineer with a Harvard MBA and experience as an Apple product marketer. He was also designer of the fast but expensive distribution system for highly regarded, but ultimately doomed, online grocery startup webvan. His experience at webvan, which went out of business in 2001, highlighted for Mountz the size and intractability of the pick-pack-and-ship problem in warehouse management. His answer to the problem was to have products come to pickers, rather than the other way around. His solution was a system that uses a large number of relatively low-cost transportation robots that can navigate, identify products, and carry out delivery orders autonomously.
Kiva robots take product orders, navigate to the shelf holding the product, and deliver it to the shipping point. The units communicate with each other to prevent collisions and avoid duplication of effort. There are two major versions of Kiva’s Mobile Fulfillment System: ItemFetch and OrderFetch. Kiva units assigned to ItemFetch bring shelves to pickers who assemble split cases into whole orders. OrderFetch units take completed orders to the appropriate loading dock door.
Mountz projects that Kiva robots could cut labor hours in half for a DC that employs 100 pickers, and by greater amounts in larger installations. In some designs, a 600-picker DC could operate with only low-cost Kiva robots and 150 workers, Mountz informed Robotics Trends in 2008. According to Mountz, the typical Kiva installation involves from 50 to 150 robots on the low end; up to 400 to 500 on the high end. Mountz further noted that Kiva’s approach is the opposite of those AGV manufacturers that might charge $150,000 per unit and require 10 systems for a DC. The same $1.5 million expenditure on a Kiva installation could involve 50 to 80 machines, each of which is simpler to direct for both pickups and deliveries than an AGV.
Kiva robots navigate using internal maps, pinpointing their location and direction using barcode stickers on the floor at one-meter intervals in a grid pattern. Each unit radios its location and direction to a central controller, which directs the units to avoid collisions. Pickers stand on the periphery of the floor, identifying the item to be picked from the Kivas that arrive as quickly as every six seconds by the laser designator the robot shines on the item.
Zappos managers claim Kiva-enabled warehouses have helped them get a product on the shipping dock 12 minutes after it was ordered. Walgreens managers say the company’s 10,000-square-meter facility operates with one-third the number of workers that would be required with a conveyor system, at a cost of about $5 million. Further, conveyor systems would cost much more and take months to construct, rather than the weeks it took to set up the Kiva system.
Though Kiva has no direct competitors in production—Kiva units are installed in more than 1,000 warehouses, according to the company—the greater flexibility and potential for additional efficiency from combined systems increases its potential. For example, walgreens, which adopted Kiva in 2007, integrates the ItemFetch with its own inventory system so that each picker puts together items for a store in the order that would make it most efficient for stockers to refill the shelves in that particular store.
Upstart Seegrid: IMR
One of Kiva’s major competitors is Pittsburgh-based Seegrid Corp., whose Industrial Mobile Robotics (IMR) vision-guidance technology is designed to allow Seegrid units to work in facilities that have not been specifically modified for robots. Seegrid’s IMR units keep detailed 3D maps of the facility in which they operate while maintaining a list of jobs to be performed. By combining information from the two sources, the IMR units use maps to guide them to the next location in the most efficient way.
Seegrid robots “learn” their surroundings during an initial warehouse drive-through. Multiple views are taken in the facility with stereo cameras or other ranging sensors, and the data are accumulated into one 3D model of the scene (a 3D evidence grid). Software breaks up the evidence grid into individual cubes called “voxels.” Seegrid units then weigh the data in those cubes to determine not only their next destination, but the best path through an aisle to get there.
Seegrid’s GT3 line of robots includes a “tugger” and a pallet mover, which requires human intervention to load the pallet but navigates autonomously. The robots are programmed to stop when they detect a human obstacle in their path; future versions will be able to avoid obstacles and take human commands, according to the company.
Seegrid has a far lower public profile than Kiva, but Seegrid’s IMR units outperform the Kiva bots in lift capacity. Kiva’s original units were able to manage loads of up to 1,000 pounds, and an upcoming version will lift 3,000 pounds. By pulling rather than lifting, Seegrid’s smaller unit is able to handle tow loads of 3,000 pounds, while its GP8 pallet truck can manage 8,000 pounds.
Though Seegrid’s customer base is smaller than that of Kiva’s, its vision-based guidance system is in some respects more sophisticated than Kiva’s. The addition of a mid-duty pallet truck expands its appeal beyond that of Kiva, whose relatively low weight limits make its units most appropriate for warehouses such as those at Zappos and walgreens, whose outgoing loads are primarily split-case—that is, made up of many products in relatively low volume per product
A guided pallet truck pushes Seegrid into the lower end of the market for customers operating wholesale DCs or distribution hubs that frequently ship orders of pallet-size or larger. Neither company offers products with the capacity for core distribution hub operators who take in goods from factories and ship them to regional distribution centers. Those facilities judge volume by the railroad car or trailer load rather than the pallet or piece, and require equipment of much greater capacity than either Seegrid or Kiva provides.
The future of warehouse automation lies with flexible, modular intelligent robotics, but it is not clear whether AGVs or flexible robotic systems will dominate the market 10 years from now.
AGV manufacturers such as Webb and HK Systems Inc. provide a far wider variety of heavy-duty materials handling equipment with guidance intelligence that makes operations simpler. New Berlin, Wisc.-based HK Systems, for example, offers a range of counterbalance vehicles with carrying capacities up to 7,500 pounds and vertical-lift capacity of 21 feet. Its unit-load vehicles come with intelligent lift decks that can roll or place a load on a dumb platform rather than requiring a crane or relying on human intervention. It also offers a range of forktruck AGVs, including one for very narrow aisles, and tuggers.
Webb similarly goes beyond pick-and-pack with an AGV rail car able to move 200 tons and AsRs systems that can be managed with the same software suite that manages its AGVs.
Though many of the heavier-duty units, including Webb’s rail car, are designed for heavy manufacturing environments rather than warehouses or distribution centers, their range of both capacity and design gives them tremendous advantages over intelligent robotic warehouse automation systems.
Currently, AGVs and intelligent, mobile robots are optimized for two different segments of the logistics market. AGVs are designed for much heavier facilities than either Seegrid or Kiva systems, which show no signs of wanting to compete in the heavier materials management space. That leaves DC hub and large-scale warehouse automation to the AGV manufacturers, even though their guidance and task-management capabilities are less sophisticated than those of the robot makers.
As Kiva and Seegrid develop the intelligence and capacity of their machines, watch for Webb, HK Systems, and other AGV manufacturers to upgrade the intelligence and guidance systems of their own units—making the heavy-duty equipment as intelligent as the lighter robots. Given the greater range of capability and proven performance, we expect to see considerable growth from AGV manufacturers into the total DC automation market by adding intelligence to their heavier-duty equipment.
Kiva and Seegrid will continue to dominate the intelligent small-package warehouse market, and will make some progress into heavier-capacity applications, though slowly. Robotics Business Review expects Kiva certainly, and possibly Seegrid, to focus on the internet commerce market—drop-ship centers, retail-service warehouses, and the like, whose requirements for high-velocity, quick-response shipping ramp the need for flexibility far above that of larger pallet- and truckload-oriented facilities.
For investors, the requirement for consumer-facing DCs that accelerate shipping and cut costs provides a great opportunity. AGV and other multisystems providers will gradually develop more comprehensive, integrated warehouse automation systems, and will dominate the market for them among the big shippers. The length of time it will take to make that happen—five to seven years before total no-human-contact integration is really practical—and the investment required to do it on a large scale reduces the potential payback.
That makes companies such as Webb and HK Systems far less attractive to growth-oriented investors than Seegrid or Kiva, both of which are entering with little competition at the very bottom of a market that will ultimately be limited, but will see tremendous growth during the next three to five years.
|The Bottom Line
Warehouse and distribution center management is chartered with increasing the efficiency of operations while simultaneously reducing costs. To address these conflicting mandates, management requires automation solutions that:
For investors and entrepreneurs, the opportunities in warehouse/DC automation are clear. Systems or companies that can provide such capabilities, based on either AGVs or intelligent mobile robotics technology, will prosper.
It is far more likely that traditional AGV providers, which have the in-dustrial and heavy materials handling background—not to mention the customers—will dominate the total-warehouse-automation market. However, up-and-coming robotic suppliers like Kiva and Seegrid fit neatly into the rapidly growing niche of consumer-facing internet commerce shippers and suppliers. This gives them tremendous growth opportunity in the short term.
For investors interested in robotics technology specifically, and for invest-ment in smart technology that disrupts some aspects of the distribution busi-ness, Kiva and Seegrid are the companies to watch.
Investors focused more on the long term, on high-capacity global com-merce and logistics, AGV manufacturers such as Webb, RMT Robotics, and HK are the better bets. These companies know how to manufacture equipment with the capacity to meet the needs of that market and have the sophistication to integrate their software with that of other DC automation and supply-chain management vendors.