Tucked away in the woodlands on the outskirts of Boston, an interesting warehouse automation story is presently unfolding. An upstart company called Locus Robotics has invented an autonomous robotics solution that has the potential to improve the efficiency and accuracy of e-commerce order fulfillment operations. What makes this story particularly interesting is that this robotics solution has been conceptualized and developed by people who are first and foremost experts at e-commerce distribution. The third party logistics firm Quiet Logistics has been working hard to enable operator-assisted automation where man and machine work closely together.
Background on Quiet Logistics and Locus Robotics
Founded in 2009, Quiet Logistics is a third party logistics (3PL) company that provides of order fulfillment services for $1 Billion of merchandise to the e-commerce channel on behalf of over 30 fashion branded companies such as Gilt Groupe, Zara International, Bonobos, Bluefly, Ministry of Supply, M. Gemi, The Fancy and Maggy London. The company operates two distribution centers totaling 459,000 square feet in Devens, MA:
- A 275,000 square foot distribution center at 66 Saratoga Blvd. in the Devens Industrial Park has a clear height of 32’. An automated picking system with 200 Kiva robots is deployed in this building. The orange robots move pods of inventory to work stations where goods are processed inbound into inventory and outbound for order fulfillment.
- A 184,000 square foot distribution center at 235 Barnum Road with a clear height of 30’. This facility handles a variety of flat pack, hanging garments and shoes from fashion companies shipping e-commerce orders nationwide. In this building, Locus Robotics has deployed a 4,000 sq. ft. test area to develop their autonomous robots.
Quiet Logistics Invests in Kiva Systems and Their Lessons Learned
Quiet Logistics was the first 3PL company to invest into Kiva Systems back in 2009. In fact the company's name originates from the fact that Kiva’s robots are quiet when they are at work. Most third party logistics firms live in a world of uncertainty because their client companies can move their distribution operation elsewhere whenever it suits their purpose. This is why it is highly unusual for a 3PL company to invest significant capital into material handling automation equipment. This visionary decision ultimately served the company well, not only from a warehousing perspective but also from a marketing perspective in that it attracted new clients.
Material handling automation enables order of magnitude efficiency and accuracy improvements that are critical to premium brand companies seeking to outsource their high-growth e-commerce distribution operations. The higher the value of the merchandise, the more important it becomes to ensure world-class order fulfillment service levels since the company's brand name and image must be protected at all times.
Quiet Logistics first started using the Kiva robots within a small self-contained 3,500 square foot area. The company subsequently expanded its Kiva System five times. Today the Kiva portion of its Saratoga Blvd. distribution center occupies 140,000 square feet, has over 5,000 shelf pods, 20 work stations, 200 mobile robots and 29 battery charging stations.
Things were going well for Quiet Logistics when the bad news hit on March 19th, 2012. This is when Amazon announced that it had acquired Kiva Systems for $775 Million. Over the next 18 months it became readily apparent that Amazon had no intention to leverage its Kiva acquisition to help other companies become more efficient. On the contrary, Amazon ultimately replaced the name Kiva Systems with Amazon Robotics and turned the company into an internal resource to modernize its fulfillment centers. Amazon has since deployed over 30,000 Kiva robots in 17 of its fulfillment centers and external Kiva clients will no longer be supported after a forthcoming non-disclosed future date. This left a handful of companies, including Quiet Logistics, with an expensive automation solution that can no longer be expanded into the future.
Bruce Welty is the CEO of Quiet Logistics and is clearly someone who understands his business. He took the time to share some of the positive and negative aspects of Kiva Systems that ultimately contributed to Locus Robotics developing a more advanced automation technology. Bruce tells of eight (8) important lessons that were learned from the company's implementation of Kiva automation.
For starters, Kiva is an expensive solution to implement and upgrade. The pods (i.e. the shelf units that are transported by the orange robots) are custom-built by Kiva and are much costlier than standard static shelving equipment. The picking/ putaway work stations are also expensive and each time a new work station is introduced, it becomes necessary to add more of the $25,000 robots to the grid. Every time the system is upgraded or expanded, it becomes a major ordeal that can take up to 6 months to accomplish from the time of order placement to go-live. In short, the addition of extra capacity to the system is by no means a simple, quick and inexpensive exercise.
Lesson number 1: The ideal automation solution needs to be simple and quick to deploy and upgrade.
From a safety perspective, the orange Kiva robots have limited capability of sensing what is around them. They are designed to work within a self-contained environment that is occupied only by machines and not people. The only way to ensure that people are safe is to ensure that they do not enter the area where the robots are functioning. At Quiet Logistics, there has been more than one occasion where one of the robots hit a person and Bruce himself has had this experience which to say the least is unpleasant.
Lesson number 2: The ideal automation solution is equipped with vision technology that enables the robot to see any obstacles in its path so that the robot can either stop or navigate around the obstacle to prevent injury to personnel.
Another issue is what I will refer to as the “black box” syndrome. This describes the situation where all inventory within the Kiva system is controlled by Kiva’s software which is a subsystem that talks to the warehouse management system. In the WMS, there is one bin location called Kiva and all transactions for that area are delegated to the Kiva control system. If anything goes wrong such as the Kiva inventory subsystem data gets corrupted for any reason, one has absolutely no idea where the goods are. This can be a terrifying thought for any company, but particularly for a 3PL who takes responsibility for inventory assets owned by its client companies. This “black box” phenomenon is not unique to Kiva since most automation systems control inventory through a Warehouse Control System. But does it need to be this way?
Lesson number 3: The ideal automation solution is designed to communicate in real time with a warehouse management system to eliminate the need for a separate “black box” warehouse control subsystem.
Another issue that is critical to e-commerce organizations, where Kiva falls short, is the ability to scale quickly and easily as peak seasonal volume increases. In a non-automated environment, one can simply add people to the operation to get more volume out the door. With the Kiva system this is not feasible since it is a human-exclusive form of automation so the system can only produce what it is designed to output. As peak season arrives, the primary methodology to ensure that Kiva does not become overwhelmed with volume is to remove items from the system and diverting them to the conventional area of the distribution center. This is impractical in an e-commerce fulfillment center where a peak shipping day can result in 30 – 50 times more units shipped than an average day. From Black Friday through to Christmas, there is a one-month time period where peak demand volumes slam the facility on a daily basis. Furthermore, it is difficult to accurately predict e-commerce growth rates from one year to the next. This level of throughput volatility is incompatible with a material handling system that is designed with rigid throughput constraints.
Lesson number 4: The ideal automation solution is designed to enable quick, easy and flexible scalability to allow the distribution center to respond to changing demand profiles and significant increases in throughput requirements.
The ineffective use of vertical warehouse space is a characteristic that industry experts associate with Kiva Systems. With the exception of Amazon, the majority of Kiva Systems are installed as floor-level operations that do a great job of managing space for the first 10 feet from the floor but all space above is wasted. Physical storage capacity is a valuable asset in any distribution center so this waste of vertical space is a definite negative. Now to be clear, it is possible to eliminate this issue by establishing multi-level mezzanines whereby Kiva robots work on each mezzanine level, but in reality this is a costly proposition. The Kiva robot weighs 300 pounds and it carries a payload of up to 1,000 pounds. This equates to a 1,300 pound weight resting on a 40” x 40” floor are or 117 pounds per square foot which suggests a costly concrete mezzanine is required to support this weight. Furthermore, moving the Kiva robots between the levels requires expensive vertical elevators since the robots need to move from one level to the next.
Lesson number 5: The ideal automation solution should be small and light enough to work on a free-standing or rack-supported mezzanine so that vertical space can be fully utilized within the warehouse.
To navigate around its designated area within the warehouse, Kiva relies on 2-dimensional bar codes that are positioned on the floor and act as guides. Thus the Kiva system needs to be tightly defined when it is implemented hence it is not the type of automation that can be deployed into neighboring warehouse aisles without a major effort. In other words, to expand the Kiva system involves a major project with a long lead time.
Lesson number 6: The ideal automation solution should be flexible enough to allow a new robot to be easily and quickly introduced to new and existing warehouse aisles without requiring major time delays or investment costs.
The underside of the Kiva robot is open and exposed to the floor such that dirt can accumulate inside its body. This requires frequent vacuum cleaning to remove dust, plastic bags, and other types of debris that accumulate within the body of the robot.
Lesson number 7: The ideal automation solution should be designed such that it can function within an industrial environment without the need for frequent cleaning.
Perhaps one of the more interesting learnings about the Kiva System is that inbound putaway labor productivity is reduced as compared to conventional labor productivity. Now this seems counter-intuitive given that the putaway process no longer involves travel time. After all, the Kiva robots transfer pods from the putaway stations into storage thereby eliminating operator travel time. How could this automated approach possibly be slower than a person manually storing the goods?
- To answer this question, the use of an example works best. Let us say that an item with 50 units needs to be stored into the Kiva System. One does not simply put all 50 units onto one incoming pod. Instead, the units need to be spread across many pods so it may be required to put 5 units across 10 pods for example.
- The reason for this is that the inventory needs to be spread across multiple pods is to minimize order picking dwell time. During outbound order picking operations, if multiple pick work stations require the same item at about the same time and the item is only stocked onto one pod, then there will be operator dwell time incurred since the pod must travel from order picking station to the next. Say for example that all of an item's inventory is housed on one pod and three order pickers require the same item at the same time then 2 of the 3 pickers may incur dwell time as they wait for the pod to arrive. Dwell time must be minimized to ensure maximum productivity so the way to ensure this is to spread each item's inventory across multiple pods.
- To quantify this statement, let us say that the standard conventional putaway process involves an operator moving the goods into the warehouse and RF scan-confirming the putaway transaction to a shelf bin. If this process can be accomplished with a productivity rate of 220 units per hour then the equivalent process of putting away the same product across multiple Kiva robots results in a comparable productivity rate of 158 units per hour, roughly a 28% loss in efficiency. Of course the pick rates are better with the automated solution, but by extension putaway suffers a corresponding loss in efficiency.
Lesson number 8: The ideal automation solution should be designed to enable productivity improvements for both inbound and outbound operations for maximum effectiveness.
Locus Robotics – An Introductory Overview
Amazon's purchase of Kiva forced Quiet Logistics to determine a new automation strategy. The outcome of this effort was the creation of a new company in 2015 called Locus Robotics funded by an initial $6 Million capital investment by a group of private shareholders. Locus Robotics now has 30 employees and the company is already working on Version 4.0 of its autonomous robotic solution. This is one automation system that needs to be seen to be fully understood. While the term “task to person” has been coined to describe how the Locus solution works, I prefer to call it operator-assisted automation.
The LocusBot is an 80 Lb. machine that can carry a payload of up to 40 pounds with its current configuration. This means that the robot can easily work at floor level or and on any existing smooth-surfaced mezzanine area. It self-navigates through the distribution center at a speed of 1.5 meters per second (~ 4.0 mph) and it can stop on a dime.
The robot is sealed off with a solid steel encasement to prevent dust and debris from entering into the interior of the machine. When not in use, the robot's battery is automatically spot charged without human intervention using regular 110 or 220V electrical outlets.
Each robot is equipped with a laser scanner that determines the environment around the robot through the use of advanced visual technology. Any obstructions around the robot are instantly detected. Upon detection of an obstruction, the robot will either try to navigate around the obstruction, or it will take an alternate travel path if this is not possible. In its next generation, a 3D visual camera will be added to improve its ability to detect the full environment around the entire height of the machine.
The user interface positioned at the top of the mast is an IPad device. The IPad is equipped with a bar code scanner to scan confirm products being picked or stored. The IPad provides information on tasks to perform (picks, puts, counts, etc).
Unlike automated guided vehicles that rely heavily on fixed markers that are placed on the floor, on the rack uprights or in the ceiling, there is no need for invasive markers or placards to be installed within the building. A standard Wi-Fi network is all that is needed. When the Locus system is first introduced to the building, the robots automatically set out to do a one-time data gathering survey of the building. The robot's cameras capture the environment as it navigates all building aisles and form an information database required to establish a digital warehouse map. The digital map is constantly updated by robots as they navigate the space.
After the initial survey is conducted, the electronic warehouse map is kept on a central server that assigns all work assignments to the robots. The primary purpose of establishing a digital warehouse map is so that the server can optimize the travel path for each robot whenever a mission is assigned. This enables overall robot travel miles to be minimized and helps to reduce the number of machines required to do the job.
The goal of Locus Robotics is to make this an automation solution that can quickly and easily scale without complexity or lengthy lead times. If the distribution center needs 5 more robots next week then simply order them online. Upon arrival, they need to be plugged in for an initial charge. Using the IPad, connect the new robot to the local Wi-Fi network and the digitized warehouse map data is transferred to the robot. Once done, the robot is ready to work. Think how Apple made our lives simpler and apply this thinking to warehouse automation and you get the picture.
How Locus Robotics Works
With Locus Robotics technology, warehouse associates are not replaced, rather their work is made more efficient. In a large e-commerce distribution center, it is not uncommon for an order selector to walk between 8 – 10 miles per day on concrete floors. For the uninitiated, this can be a brutal assault on the knees and ankles. Much of this travel time can be eliminated by changing the work process. Traditionally this has been achieved by implementing conveyor systems that move goods from the point of picking to sortation systems or packing areas. The negatives with powered conveyors are that they are costly to install and maintain; they may require downstream sortation systems; they can be noisy; they can be difficult to change once they are bolted down to the floor; and they typically end up blocking cross aisles and dock areas or somehow creating travel disruption.
An alternative approach is let the robots do the horizontal transfer work instead of the conveyor system. This enables the operators to remain in the operating aisles throughout the day. Particularly for distribution environments where orders are small, the Locus Robotics solution provides a viable alternative to conveyors and other forms of horizontal transfer systems.
As outbound orders are planned and released by the WMS, the WMS generates pick tasks. In a conventional environment, operators with RF devices would then pick and confirm the orders. With the Locus solution, the pick tasks are assigned to the Locus Robots for the area that the robots work within. Each associate is assigned to a zone within the area and he or she collaborates with the robot on each task. As work tasks are generated and prioritized by the WMS, the tasks are assigned to the robots. The Locus software determines how best to dispatch the machines so that they take the shortest travel path to move to the location where the inventory transaction takes place. As the robot arrives at the bin location, the IPad device displays the work task. The operator walks to the bin location and performs the task. The item's bar code is scanned by the IPad to confirm that the correct product is being handled. In the case of order picking, the operator deposits the product to the tote to complete the task. The robot then moves on to its next destination, which could be another pick location or it could return to the packing area.
In short, this approach has the operator patrolling an area looking for robots that are stationary within the operating aisles. When the robot's lights are yellow and flashing then an urgent order is awaiting a pick task, otherwise the green lights flash for regular orders. The same concept applies to putaway tasks since the robots are used to transfer inventory from the point of receiving to their putaway location in the warehouse. The fact that putaway and picking tasks can be performed together enables real-time task interleaving which has the potential to further improve labor efficiency.
Recap of Benefits and Looking Ahead to the Future
Any critical thinker in the distribution industry will quickly point out all of the things that Locus does not yet do but it is important to put everything into context. This is a solution that was first deployed in a warehouse environment in March, 2015 and the complexity of having man and machine work closely and safely together in close proximity requires a great deal of brain power. Furthermore, the software that is required to optimize the workflow and travel patch of the robots is far more complicated than the casual observer can appreciate.
Did Locus Robotics manage to address the eight learning lessons discussed earlier within this article? Indeed this is an automation solution that is easy to install and deploy and it is flexible enough to scale quickly and easily. It is safe and does not require a human-exclusive environment. It does not require its own “black box” subsystem to manage inventory. With an all-in price tag of approximately $35,000 per robot, this is an automation technology that allows companies to crawl, walk and then run rather than investing in a monolithic solution on that breaks the bank out of the starting gates. Lastly, it enables productivity efficiency for both inbound and outbound operations and can be deployed on mezzanines to enable the full use of vertical warehouse space.
Having said all of this, there remain opportunities to improve the Locus Robotics automation, some of which are proposed below. We ran these suggestions past Locus Robotics and include their feedback below.
- Integrate a weigh scale underneath the tote to enable weight capture at the point of picking. This way any quantity mispicks can be detected at source rather than being captured downstream at the packing tables. Since quantity errors tend to make up the majority of mistakes, this capability would help to improve accuracy to near perfect levels.
It is great that putaway and picking tasks can be interleaved but rather than having operators patrolling their designated work area looking for robots, have the operators be instructed to go to the next location where a task is to be performed. Introduce an Apple watch or some other cool gadget that displays the next say 10 tasks to the operator. This would reduce time loss caused by the operator always being on the go moving to the next pick or put location.
- LR: The use of integrated weigh scales will add cost to the system and may or may not catch quantity errors for light-weight products if they fall within the acceptable tolerance percentage so this option may not necessarily be beneficial to all operations.
Design a robot that can carry more than one tote so that the machine can move in and out of the storage aisles less often and more tasks can be executed for each mission.
- LR: Locus Robotics has already simulated "perfect knowledge" vs. "random patrolling" and to their surprise found that there was no material gain in productivity if a worker knows where the next task is. Having said this, maps / notifications are being contemplated to help ensure that priorities / service levels are managed at all times.
Display an image of the product being handled on the IPad so that the operator does not confuse the unit of measure and pick an inner pack instead of a retail unit.
- LR: Locus Robotics has already demonstrated the ability to carry more than one tote by just clipping in another shelf onto the mast as well as the ability to add a large storage platform to the top of the robot. The mast has the flexibility to hold specially designed armatures / accessories. As well, a storage armature that can hold 200+ Lbs of merchandise will be added to the mast in the future.
Enable work areas to be dynamically assigned to operators so that workload balancing can automatically be determined by the system. A side-benefit of zoning operators is that they become responsible for the own work areas. Dynamic zoning reduces this benefit but it is an important requirement for any warehouse where tasks are carved up into multiple work areas.
- LR: If the operator touches the IPad screen anywhere, a picture of the SKU will appear thus this functionality already exists today.
- LR: Locus Robotics agrees that this idea is great and is part of their design. The expectation is that every implementation will require some specialized optimization. These optimizations will ultimately appear like "apps" on an iPhone and will be available to clients. These optimizations will likely include systematic zone definition, dynamic robot and worker assignment and other ideas. The Optimization Library is currently being built and will likely be endless in its potential ideas.
Why would a company invest in the Locus Robotics automation solution? What environments is this solution most applicable? What are the benefits?
In our opinion, the Locus Robotics solution is best suited to e-commerce fulfillment operations or other types of distribution operations that have a high quantity of small orders to process for a wide variety of SKUs spread across a large building footprint. Having autonomous robots performing the horizontal travel work can increase picking and putaway productivity work by a factor of 5-8X (ed. Note: this productivity increase factor will be different for each operation and is relative to the base productivity rate of the conventional operation) for work tasks performed at floor level. This type of automation will be of interest to distributors with unit picking environments that require the flexibility and scalability to handle peak seasonal volumes that can be significantly higher than normal volumes.
The payback for this type of solution will be derived from the ability to move more volume out the door without increasing head count within the distribution center or similarly, the ability to ship existing volumes with fewer people. Either way, the ability to leverage a solution that is not tied to the building, such as a mezzanine or a conveyor system, should have great appeal to any company that has an uncertain future in terms of growth rates. If there is a need to move to a larger building then technology can easily be moved over a weekend without disruption to the existing business. As well, companies seeking rapid deployment in greenfield and/or brownfield distribution centers will be well-poised to leverage the ease of set-up and the quick buy-to-deploy cycle which is expected to be in the 4 to 7 week implementation timeframe (WMS integration being the critical path element).
While this solution is still in its infancy, we believe that Locus Robotics is definitely moving in the right direction. We are also strong believers in the future of autonomous robots working within order fulfillment centers, particularly in the rapidly growing e-commerce sector.
Marc Wulfraat is the President of MWPVL International Inc. He can be reached at +(1) (514) 482-3572 Extension 100 or by clicking here. MWPVL International designs conventional and automated distribution centers and material handling systems and can help your firm evaluate the most appropriate solutions for your manufacturing or distribution operation.