A little over ten years ago the first voice directed bucket brigade started up in the Deerfield Beach, FL distribution center of Republic National Distributing Company. As the name suggests, in bucket-brigade picking workers pick items into totes or cartons and pass the carton along to subsequent workers in a pick module. Bucket brigades use a zoneless, pick-and-pass process which creates a highly efficient and self-balancing work unit that eliminates bottlenecks without direct supervisory intervention.

There’s a great description of the bucket brigade process on the Web, including a short animation (thanks to John Bartholdi and Donald Eisenstein). We also have several good live examples of Jennifer voice directed bucket brigades in action, including this clip from a wine and spirits distribution center. In this video, the picker has the dialogue working at top speed – she’s picking hundreds of bottles an hour. Jennifer tells her what shelf location to pick from and how many to pick. The picker speaks the location check digit and quantity. She says “pass carton” to pass the carton to the next picker in the module and then enters the carton ID from the next carton in line. Lucas has dozens of other examples of bucket brigade pickers in a variety of DCs (contact us if you’d like to see other videos).

Bucket brigades are nothing new in the warehouse, but before Jennifer was installed in Deerfield Beach, bucket brigade picking systems relied on paper pick lists that traveled with the cartons down the conveyor. The problem with paper is that it creates wasted time while pickers read and handle pick sheets. So some DCs installed pick to light systems. While lights provide productivity, they are costly to install and maintain. And some pick to light picking systems use fixed picking zones – these are not bucket-brigades. Restricting each associate to a fixed number of bays defeats the purpose of the bucket brigade.

And that’s where Jennifer comes in. Republic National Distributing Company saw an opportunity for voice in their bottle pick module, which had been using paper. So in 2001, the first Jennifer voice-directed bucket brigade was fired up. The system delivered high double-digit productivity gains along with reduced picking errors. Based on the initial success at RNDC, Jennifer spread like wildfire in the wine and spirits industry, where she is now used at dozens of DCs.

The story doesn’t end there. Retail and other industries also started looking at voice as an alternative to their expensive light systems, but the conventional wisdom was that voice technology would slow workers down compared to light-based picking. Through innovative dialogues and other process changes Lucas has been able to deliver Jennifer applications providing equivalent or better productivity than light-based bucket brigades at a number of major retail DCs. In fact, Jennifer has replaced light-based picking at a number of retail fulfillment centers.

Compared to the pick to light systems she replaced, Jennifer also delivered improved accuracy, in large part because pickers are required to explicitly confirm the quantity picked – they say “grab two” when picking two pieces. Voice-based bucket brigades also provide complete accountability since the system tracks who picks what into which order carton. To get the same level of accountability with a light-based system, associates would need to pick in fixed zones, which will inevitably lead to idle time and defeat the productivity goals of the bucket-brigade concept.

One of the process innovations in voice-directed bucket brigades in retail, publishing and other industries, is the ability to batch orders in a train of cartons, so that the picker reaches into each slot once to fill multiple orders. The first Jennifer batch bucket brigade system was installed as far back as 2004. These batch picking systems have led to ultra-efficient picking processes that just aren’t possible with lights or other technologies. Compared to traditional single order bucket brigades, batch picking bucket brigades offer additional incremental productivity gains. Not the huge 20 percent or greater gains you get from moving from paper or scanning to voice, but measurable gains resulting in tangible savings. Best of all, these additional points of productivity are free – they are the product of creative process engineering rather than additional capital investments.

In voice picking - and a marathon - a pace-setter can push the field to run faster.

Last month a few folks from Lucas participated in the City of Pittsburgh Marathon. Our team included one marathoner, two half marathoners, and a team of five relay runners (slackers!). The interesting and relevant news is that the overall winner of the 26.2 mile race was the runner who was hired to set a fast pace – the pace setter or “rabbit.” The rabbit’s job is to push the other runners to run a fast race.

So what does any of this have to do with voice picking?

Pet Supermarket, a regional retailer based in Ft. Lauderdale, has its own Rabbit, according to Oscar Martinez, Director of Distribution. Oscar knows that when he assigns this guy – let’s call him Roger – to pick in the “dog pound“ (a pallet rack section where they pick heavy bags of dog food and similar items), Roger is going to do his best to pick faster than the guys who typically work there. And the other guys in the dog pound are going to put out a special effort to beat Roger.

At break times the guys check their productivity rates with managers who can view up to the minute statistics using Jennifer’s Engage Management Console. (Engage is the Web-based management reporting, configuration and administrative application that is a key part of Jennifer VoicePlus.) At Pet Supermarket this friendly competition boosts productivity. But it’s only possible if there’s a way for users to get feedback on their pick rates throughout the shift.

Likewise, objective performance feedback is a key component of any well-designed labor standards program. If you want your workers to meet their goals, they really need to know where they stand. The problem for many labor standards programs is that the feedback comes after the fact – at the end of the shift or the end of the week, when its too late for workers to pick up the pace. That is, until now.

Lucas systems recently introduced a Speedometer module to address this challenge. Speedometer gives Jennifer voice users real-time productivity updates as they work. So while a user is working he can ask “Jennifer what is my current productivity?” Better yet, DC managers can configure Speedometer so that Jennifer will warn a user if his or her productivity drops below 90% of pre-defined pick rates (based on historical averages or engineered labor standards). The first customers that rolled out Speedometer have seen immediate gains in productivity above and beyond the increases they experienced when first moving to voice. In each case, Speedometer was turned on after users had been working with Jennifer, at a point where the initial productivity gains from moving to voice had reached a plateau.

The direct productivity gains attributed to Speedometer are typically a few incremental percentage points above what is gained in moving from RF or paper-based picking to voice. And in today’s hyper-competitive business environment, a few percentage points is what separates a winner from an also-ran.

If you read  logistics trade magazines you’d think that multi-modal applications combining voice and scanning are sweeping the industry. Despite the buzz, many people I talk to still aren’t convinced that scanning has a place in a voice picking process.

I saw a great example of where scanning does fit with voice when I visited Oriental Trading Company earlier this week – dodging the blizzard along the way. OTC, a leading direct merchant, is using Jennifer for picking in a highly automated fulfillment center in La Vista, just outside of Omaha. OTC uses an “explode and assemble” fulfillment process. Deon Wagner, Director of Warehouse Operations at OTC, did a great job breaking that down for me: they explode incoming orders into waves and sub-waves for picking, and then they re-assemble the picked products for shipping using miles of conveyor, sophisticated sorters and WCS software from W&H Systems.

For picking, OTC is using Jennifer in a variety of processes, including a case-break application in which full cases of product are delivered to associates who pick and put to totes in a pick module. Here’s how it works.

Associates first set up totes in 21 locations on pick module shelves. They scan each tote barcode and then the barcode on the corresponding module locations to associate each tote with its position in the module. (This tote identification is needed by the downstream order re-assembly systems.) Associates could set up the totes using voice, but when you observe the process its pretty clear that they can scan a lot faster than they could read all the tote and location check strings.

After setting up the totes, the associate is ready to start pulling product. To begin, she scans a barcode on a carton of product, opens the case and scans a barcode on the first item she takes out. This product scan is to confirm that the right product was pulled from reserve. Associates have the option to read the last 4-5 digits of the carton label or item barcodes, but everyone I saw that day was scanning. I asked one of the workers why. She said, simply: “Its faster.”

From that point forward the put-to-tote process is all voice. Jennifer tells the associate how many items to take out of the case and then how many to put into which tote: put three in the tote in the red two position. The associate then confirms what she is doing by reading the check digits on the shelf labels: “zero two [location check digit], put five [quantity].”

Seven or eight years ago, OTC probably would have done this whole process by voice because there were no voice-capable mobile devices that included a scanner. In those days, a voice-plus scan process would have required a voice-only terminal and an additional scanner for every user. The extra cost of the scanner might not have been justified by the time savings of scanning. But since OTC is using Jennifer on standard Motorola MC9000 multi-modal terminals with scanners, they get the benefits of voice plus scanning at no additional cost.

At OTC the advantages of scanning plus voice are clear. They get a few minutes of time savings per user per hour, and those minutes quickly add up, especially in busy periods where every minute counts. And there’s no cost to get scanning and voice in a single system – it’s a standard capability of the Motorola hardware and the Jennifer VoicePlus software.

To learn more about OTC’s use of voice picking with scanning and other automation systems, consider attending the NCOF conference in Las Vegas this April.  Deon Wagner of OTC will be presenting a case study on how his DC uses Jennifer and other technology to ensure on-time, low-cost distribution. Attend Deon’s presentation on April 5, and stop by to see Lucas at the show.

Since 2004, the average price of a mobile computer that can be used for voice picking applications has dropped by more than 60 percent. Back in the early 2000s, the only option warehouse managers had was to buy an expensive, single-purpose voice appliance offered by one of three hardware manufacturers (some of which were perpetually skating on thin financial ice).

In 2005 Symbol Technologies changed the market forever by introducing the first voice capable MC9060 terminal. The MC9060 was originally designed as an industrial workhorse primarily for scanning-based applications in warehouses, stores, and industry. Working with Lucas, Symbol (which was acquired by Motorola in 2008) upgraded the audio capabilities of the MC9060 and created the industry’s first true multi-modal mobile computer – a single device that combined voice, scanning, touchscreen and keypad in one.

Since then all of the other major mobile computing vendors have introduced voice-capable devices, including more powerful rugged devices, compact wearable terminals, and smaller, lighter, and lower-priced handheld models. So now, instead of spending $4K or more for an “appliance,” DCs can buy multi-function voice-capable computers for less than half that cost. Just like in consumer electronics, warehouse mobile computers are getting more powerful and less expensive every year.

Dropped your Droid? Replacing it is still better than buying a rugged mobile computer. In theory, at least.

Still, compared to a $100 or $200 smartphone, a $1,500-$2,000 mobile terminal seems expensive. Not surprisingly, every week we talk to DC managers and IT folks who ask if they can use a $100-$200 smartphone, PDA, or even an iPod Touch for voice in their warehouse.

There’s good logic for why you might want to use a dirt-cheap voice terminal. First and foremost, with a low-cost hardware option, initial investment costs for voice systems are dramatically reduced. As for reliability concerns about using consumer-grade devices in a warehouse, the argument is that if the device drops and breaks, you can just replace it. In fact, at $200 a pop, you can replace each device several times a year and still have a lower hardware cost. (In reality, the cost to buy a smartphone is probably closer to $400 or $500 if you aren’t buying a cell plan. But that’s another issue.)

This all seems to makes perfect sense. Except it doesn’t. Here are two reasons why.

• Noise Reduction. Warehouse voice systems have to deal with loud noises that change without notice: lift trucks, horns, conveyors in the background (turning on and off), pallets banging to the floor. Filtering out background noise is one of the great challenges of voice recognition in a warehouse. It takes noise-cancelling headsets, and additional noise filtering technology on the mobile device to handle it. Unfortunately, mobile operating systems used on smartphones (including Android) don’t provide a means to adequately filter and regulate noise levels, so they don’t provide acceptable voice recognition rates in noisy environments. (A typical warehouse voice application provides 99% accuracy or better, so users rarely have to repeat themselves.)
• Battery Life. Cell phones, smart phones and PDAs don’t typically do a lot of compute-intensive work. Voice recognition puts a real tax on processing power and batteries, which is why the battery of a standard industrial computer weighs as much as a cellphone – they have to last for a full shift. With a consumer device users would probably have to change their batteries every hour or so. And if you want to use an iPod, you better have 4 or 5 devices per user since those batteries don’t swap out.

The bottom line is that we’re not likely to see a viable $100 voice terminal for years. But we’re starting to see some multi-modal devices and rugged PDAs with street prices of less than $1000. Compared to prices 5 years ago, this is a dramatic shift. And these devices offer far more functionality than yesterday’s voice-only appliances.

Lucas Systems recently introduced a new voice recognition platform – Serenade – that combines the advantages of speaker-dependent and speaker-independent voice recognition systems in a single product. Besides industry-best recognition accuracy in challenging warehouse environments, with Serenade users can start using the voice system after completing a short enrollment process, as compared to a 30 minute voice training process required with traditional voice picking systems.

To understand the advantages of Serenade, its useful to look at the pros and cons of speaker-dependent technology that dominates the warehouse market today, as well as of speaker-independent technologies that are more prevalent in consumer, call center, and other markets.

Consumer speech recognition is a rich target for comedy. See some funny speech-related videos on the Lucas YouTube channel.

Dependent Systems Require Training and Re-training

Until recently warehouse voice systems used speaker-dependent technology in which the voice engine is tuned to recognize each user’s speech patterns. Those systems required a 20-30 minute “voice training” process for each user. This half hour training step is seen as a small price to pay to get near-perfect recognition accuracy. One shortcoming of most first generation speaker dependent systems was that they often required users to record a second voice template after starting to work with the system. Why? Because people usually speak very clearly and deliberately when they initially create their voice template, but as they get comfortable working with voice, they speed up and revert to their usual speech patterns. When that happens, the recognizer starts having problems matching what they say against the template they built, so the user will retrain.

To eliminate the need to retrain, three years ago Lucas introduced the concept of adaptive voice modeling, in which the speech software automatically adapts the user’s voice template in the course of use. So as users start working faster, slurring words together and cutting off the ends of words, the recognizer keeps up. More importantly, rather than degrading, recognition accuracy actually improves with use, even when a user comes down with a cold.

Independent Systems Don’t Adapt Well to Individuals

What if you could do away with training altogether so that anyone could use the system without creating a voice model? That’s the advantage of speaker-independent systems used in automated customer service applications and other consumer-oriented products. Unfortunately, speaker-independent systems did not provide high accuracy rates in challenging industrial settings, which would force users to repeat themselves or, worse yet, frustrate users to the point where the system becomes unusable.

Another common drawback to speaker-independent systems is that they may have trouble with non-standard speakers – people with heavy accents or speech impediments. And since these systems are not designed to work with user-specific models, they have no way to adjust to challenging users and individual speech patterns.

Serenade Gives the Best of Both Worlds

Serenade provides the advantages of minimal training and adaptation in a single product. What that means is you save 20 minutes or more in initial training, get high accuracy from day one across all users (even challenging users), and don’t have to re-train after getting comfortable with the system. For a DC with a couple of dozen users, every 20 minute time savings equals one eight-hour work day. And in today’s economy every little bit of time and cost-savings matters. More importantly, with industry-best recognition rates, users get confidence in the system from day one and can concentrate on their jobs rather than the technology they are using to do it. At the end of the day, that’s the critical requirement for any voice system.

Last  week my  laptop crashed leaving me without a PC for a couple of hours. My first thought: “How am I going to write anything without a PC?” Fortunately, I was able to keep up with email on my iPhone (I can’t imagine having a phone that is just a phone!).

This PC-free morning got me thinking about single-purpose word processors, which were just hitting the streets when I was in college, but which have long since been added to the trash heap of late 20^th century technology. There’s a direct parallel between word processors and today’s single-purpose hardware used for voice directed picking and other warehouse applications. Single purpose voice-only devices have been around for years, but its just a matter of time before they go the way of the word processor (and single-purpose cell phones, for that matter).

Typewriter to Word Processor to PC

I entered college in the early 80s with a portable Smith Corona electric typewriter and ended my academic “career” typing my senior thesis on an IBM- compatible PC in the college computer lab. Some classmates owned word processors, which were a cross between a typewriter and a computer, with a tiny screen. PCs were still expensive and required an external printer, and I needed multiple floppy disks to hold my senior thesis. But typing in the computer lab was a lot better than the old way of hand writing a draft, and editing as I typed with a trusty bottle of white-out at the ready.

Fast-forward two decades or so and every college student has a PC, my iPhone is hundreds of times more powerful than the PCs in a mid-1980s computer lab, and the single-purpose word processor is a relic of the past.

 Voice-Only Terminal To Multi-Purpose RF Device

Now compare that to the hardware used for voice directed warehouse applications. Thirteen years ago, when Lucas was founded, all of our applications were delivered on single-purpose voice-only terminals that were made by specialist companies. But when Motorola introduced its first voice-capable RF terminal in 2005 (with assistance from Lucas, by the way), we were quick to deliver our applications on the MC9000 and other multi-purpose mobile devices from Motorola, Intermec and LXE.

Lucas Applications were originally delivered on simgle-purpose voice-only terminals like the ones in this picture.

Today, DC managers considering voice applications can use the same hardware they use for RF applications, instead of buying new voice-only terminals. And voice software vendors like Lucas have introduced powerful new ways to leverage the scanner, touch screen and other capabilities of multi-purpose hardware alongside voice to drive improved warehouse processes. So along with a lower total cost of ownership, multi-purpose terminals deliver bigger benefits for warehouse operations.

Just like the word processors of my college days – and the talk-only cell phones of the recent past – it’s just a matter of time before we look back on single-purpose voice terminals as a quaint reminder of how far warehouse technology has evolved. The only question is whether that happens in one year, three years, or five?

Most DC managers view temporary workers as a necessary evil: they need the extra hands to meet seasonal order peaks, but they dread the extra work required to bring temps up to speed, not to mention the inevitable increase in errors from inexperienced workers. John Migliore, General Manager of the Do it Best Retail Service Center in Montgomery, NY, is an exception.

I met John earlier this summer at a voice-directed DC Tour event Lucas hosted in Montgomery. John’s team had just completed on-boarding a couple of dozen temps for Do it Best’s busy summer season. Prior to the event John gave me a tour of his facility and told me how the Jennifer voice picking system has benefitted managers and associates. Among other things, he said training temps is no longer the challenge it used to be in the pre-voice days. (To see the Do it Best voice application in action, view the video on YouTube: Voice Picking at Do it Best)

On-boarding temps takes time from supervisors, team leads, and other pickers. Do it Best still needs to provide general orientation and safety training, but John says the specific picker training is a lot faster and easier with Jennifer. And since many of Do it Best’s summer temps are college age people, they like the idea of using voice technology and a mobile computer. After all, most 20 year olds have a smartphone permanently attached to their hand.

John says that with Do it Best’s old paper picking system it would take at least a week to train a temporary worker, and  4-8 weeks before they would reach their target performance rates. With the Jennifer  voice system, most pickers are up and running on their own in one day and up to standard within a  week or two. Beyond the short learning curve and productivity gains, John also explained that the voice system, since it validates locations and quantities at every step in the process, has greatly reduced the number of errors they see from temps. John was so confident in his temporary associates that we mic’ed them up for the tour.

While most DC managers realize voice will improve the accuracy and productivity of their permanent workforce, few people appreciate that those benefits are magnified when dealing with inexperienced workers, especially temps. This is an important message as we approach the holiday shopping season, when many other retailers and foodservice and grocery distributors face their own seasonal peaks.

A Lucas project manager just returned from a Jennifer voice-picking deployment with a story of “Joe The Selector” who thought voice was never going to work in his DC. Like most other voice skeptics, this “Joe The Skeptic” became “Joe The Voice Champion” his first day working with Jennifer.

Joe is not alone. Lots of new users view a voice deployment with a mix of technophobia and skepticism.

Technophopia

Warehouse workers are no more fearful of technology than the rest of us. But DCs moving from paper to RF or voice have to take special care with users with limited computer experience. Fortunately, moving to voice is typically easier than moving to RF. While RF apps often use function keys and special input codes, voice applications rely on verbal instructions and spoken responses that are short, simple and intuitive (click here to access videos of voice applications in use). Better yet, today’s modern voice systems running on standard RF terminals like the Motorola MC3190, Intermec CK3, or LXE MX8 offer easier, more user-friendly  training and sign on processes than legacy voice systems using voice-only terminals.

Voice picking user log on is as easy - and secure - as using an ATM.

Signing on to Jennifer on a standard RF terminal is like signing on to an ATM – just touch a button on the screen, and key in a username and password. (This screen-based process is also more secure than a voice-enabled process, but that’s a different subject.) Likewise, voice training with Jennifer uses audio prompts and screen-based information (text and pictures) that guides users through the process. This is a far better way to introduce people to their voice terminal than using a special-purpose training device (which may be required with some voice-only systems) or a training program that uses audio-only or text-only prompts.

 Skepticism

When some people hear they are going to be using voice recognition systems, the first thing they think about is a bad experience with the voice recognition systems (IVR) used by credit card companies, banks, insurance companies, etc. “If a voice recognition system can’t understand me when I’m talking on the phone in my living room, how is it going to understand me when I’m picking cases in my warehouse?”

The point is that you wouldn’t want to put an IVR system in a warehouse.  While the underlying technology used in those general purpose systems has advanced tremendously, warehouse voice applications require additional recognition and noise-handling capabilities to work effectively. That’s why it’s so critical to use voice technology that is proven in a warehouse. (For more on this issue, see an earlier article that looks at how recognition impacts user acceptance.)

The bottom line is that today’s voice systems offer built-in advantages to help overcome skepticism and technophobia. But the technology only goes so far: The surest way to turn Joe The Skeptic into a champion is to get his buy-in before you install the system. That is the subject for a future post.

Do you have anything to add to this subject? Whether you’re a skeptic or a technophobe, we’d like to hear from you. What strategies do you use to get associate buy-in for process changes or technology upgrades in your DC?

Is this really any different than texting-while-driving?

With the rising numbers of texting-related driving accidents, more and more states are banning texting while driving (to see the laws in your State, use this link).  But think about this: is texting while driving really any worse than using paper pick lists or an RF terminal in a distribution center, especially if you’re “driving” a pallet jack or fork lift?

So maybe it’s time to bring the same thinking to the distribution center floor, where associate distraction is a leading factor in lost-time injuries, in addition to picking errors and lost productivity. The answer: put down the terminal, put down the pick lists, and let Jennifer™ guide you by voice.

We All Do It
I’ll admit it, and I’d guess that most of you would admit to being distracted by a cell phone in the car, or even texting while driving.  At the very least, we’ve all seen someone else on the road making bad lane changes or blowing through an intersection while holding a phone to their ear or in their hand!  Put a hands-free headset on the phone, use voice recognition to dial calls, stop reading your texts and emails, and you’ve improved your concentration on the driving task, making the roads safer for everyone. (For another view on the subject of texting-while-driving, see Mitch MacDonald’s commentary in the April issue of DC Velocity magazine.)

Now think about the distribution center associate who is juggling paper lists, pens, and/or RF terminals, looking for locations, looking for correct items, and trying to remember the correct quantity to pick, in some cases while maneuvering a cart, pallet jack, or other equipment.  Compare the eyes and hands-free driving experience to the life of the DC picker and you’ll start to appreciate how voice direction and voice recognition improve associate safety.

Quantifying the Safety Benefits?
Do it Best’s DC in Waco, Texas has a big banner hanging out front touting 2+ years since a lost-time injury. Brent Watts, the DC General Manager, says the Jennifer voice system is a big reason for their spotless safety record. It’s hard to quantify the direct dollars and cents safety benefits of voice since there are lots of factors that contribute to safety. Nevertheless, there’s no doubt that voice makes for a safer workplace, just as texting-free roadways are safer roadways.

What are your thoughts? Is paper-based order picking really as bad as texting while driving? Submit a comment to join the conversation.

Last week Do it Best Corp. opened the doors of its Waco, TX Distribution Center to folks from other warehouses in the area. Do it Best has a great success story to tell (use this link to see a video case study about Do it Best), but that wasn’t the most striking thing about the tour. What struck me was how surprised most of the tour attendees were by the depth and richness of the Do it Best System compared to legacy voice systems they had seen at other DCs.

The fact is, most operations people today have a general understanding of voice, since its been around for years and is no longer a true “bleeding edge” technology. But when you get beyond the voice-picking basics, it’s amazing the degree of flexibility available in today’s voice applications. There are an almost limitless number of ways to handle exceptions, configure voice dialogues for different users, and combine scanning with voice. And beyond the hands-on user capabilities, today’s voice applications also provide incredibly rich management reporting and process management tools that help supervisors do their jobs better. Long story short, if you haven’t seen a modern voice system in use, you might be surprised by what you see.

We will be posting the dates and locations for upcoming Voice-Directed DC tours on this blog. So if you’re interested in taking part in a future event, and seeing Jennifer up close and personal, sign up for blog updates using the box at the top of this page. (Or fill out our request form and we’ll add you to our mailing list.)

How do you feel about hosting or attending warehouse tours? Are DC visits more valuable than trade shows and conferences? Or do trade shows and tours serve different needs: education versus networking, etc.? Please submit a comment to join the discussion.