Voice Control Powering Distribution Efficiency

Back in the 1980s, I was on a supplier audit mission and requested a review of a distributor's warehouse system. I was taken to the back of a very large building, guided through secured double doors, and introduced to the unexpected.

Above me was a system of operating gantries with picking claws carrying various colored component bins to some final destination beyond my field of vision. When I looked closer, I saw that the gantry network branched into various “aisles in the sky.” I asked my guide to explain how the system worked, and he was more than happy to show off this new technology.

I followed him to a computer screen where he punched in a part number and said, “Watch this.” He directed my eyes to the hanging rail system to watch an empty mechanical claw move towards a bin location about 8 feet above my head and down the aisle about 20 feet. I watched the claw stop in front of a bin, extend forward, pick up the bin, and move it to a receiving station, which I could now see. When we arrived at the receiving station, the bin had been lowered and set on a rack with a paper label that matched the computer printout being printed at the receiving station as I watched. I felt I was on the set of a Star Trek episode.

We have come a long, long way since then. Now it is a common sight to see distribution centers with picking apparatus enabled by both computer and voice commands via a wireless network. The picking operation is effected by someone with a headset terminal for hands-free operation. An assignment is transferred from the warehouse management system via an 802.11-enabled headset or terminal. The voice-enabled device translates the assignment data into audible commands. The user provides spoken responses to confirm any actions, and the confirmed instruction is converted to data that the host system can use to self-update.

Voice technology has now been deployed at thousands of companies around the globe in multiple industries, for multiple workflows and processes, and has moved well beyond the initial picking applications that were the first areas of deployment focus in the distribution center.

One voice technology vendor, Vocollect claims 500,000 users. Voice tech has only been around for about a decade, and for one vendor to have half a million users is a testimony to the perceived value of this supply chain innovation.

Off-the-shelf software with pre-populated phrasing tied to text commands is fundamental to rapid deployment. In the most basic terms, a voice system directs an operator to complete discrete tasks, such as an order-pick or replenishment, using audio commands. That generally means that the “text-based” commands originating in a warehouse management or order processing system have been converted into voice commands by the voice technology provider. The software continues to evolve with each new generation, and so the implementation for voice-controlled systems is becoming easier and therefore more desirable. The reality today is the success rates for voice technology are very high, and “failures” are virtually unheard of.

Voice today is being used in put-away, truck-loading, receiving, inbound quality/compliance audits, and really just about every process in the distribution center. The industry may soon be able to use voice command for offloading trucks and containers equipped with the mobile picking apparatus. It shouldn't be too long before we see beverage distributors offload case after case of bottles and cans to retail stores and vending machines via voice control operations.

Can you hear it? “Give me 16 cases of Mountain Dew, 4 cases of Pepsi, and 1 case of 7-Up.” The side-accessible delivery truck, equipped with a computer-controlled moveable gantry with mini-forklift tongs, moves up and down and left and right and pulls the heavy cases and lowers them to the dolly to be wheeled to the in-store final destination. The packing slip shows up on a portable touch display that receives the accepting signature and a copy of the invoice beams through the cloud to the store's AP department.

Because this technology does not depend on the machine-based picking system, an employee could be a full time “picker.” Equipped with a headset, the picker receives voice command for pulling batch orders. When the goods are pulled, one component at a time and individually voice confirmed, a hand-filled bin is hand-carried to its final destination awaiting the customer's retrieval.

I almost engaged a grocery service that used this system for batch orders via phone. I decided that maybe I didn't need another technology that would make me even more sedentary than I already am. But for those big distributors handling thousands of orders a day, how could you not love this system? This is one innovation that really speeds up the supply chain and improves accuracy via voice confirmation one SKU at a time.

To Vocollect and other voice system purveyors, I say “Thank SKU very much.”

13 comments on “Voice Control Powering Distribution Efficiency

  1. Cassandra @ Voxware
    September 20, 2012

    Voice technology is rapidly evolving to provide organizations with more functionality, make systems and task more efficient, and to drive profitability. To keep up with demand and to continue the evolution of voice, Voxware recently introduced a cloud-based voice solution, cloud voice management suite (cloudvms).

    Cloudvms empowers organizations, especially those who typically could not undertake a on-premise voice solution, to enhance operational productivity, improve accuracy, and better serve their customers without investing in costly infrastructure. To learn more visit,

  2. Barbara Jorgensen
    September 21, 2012

    Fascinating, as Mr. Spock would say. I particularly like the idea that the voice command is turned into text. My main question was how records are kept because heaven knows we love our records!

  3. bolaji ojo
    September 21, 2012

    Douglas, Voice technology must have applications in other manufacturing areas of the electronics industry. How is this applicable for other manufacturers?

  4. t.alex
    September 21, 2012

    In the manufacturing area, I think it's a bit tough due to the noisy working environment.

  5. stochastic excursion
    September 22, 2012

    Voice recognition technology has been around for quite a while and the wizardry involved has given rise to some remarkable capabilities.  Software being pioneered today for air-traffic control and other aviation applications can filter out a single speaker from a crowd, even where the speaker's voice is affected by stress or illness.

  6. Mr. Roques
    September 26, 2012

    So the company only provides the technology for the voice recognition? not the automation? I can see how not having the right coordination might bring a big set of problems.

  7. dalexander
    September 28, 2012

    @Bolaji, Recently I have been following a company called “Rethink Robots.” They have a product being released now that can take instruction without a software program direct application. By showing the robot what to do, the robot, Baxter, learns and repeats the action and hones its skills as it is learning. With voice to text technology, I can see where it won't be long before this technology will merge with robots that could be assigned to flexible assembly lines. Japan has a series of robots that respond to human voice prompts and so it is possible already for a robot to “hear” a voice command and respond with a physical gesture. With super computers handling billions of instructions per second, how soon until we see AI in assembly lines is a question that will soon find an answer. Voice responsive robotic movements should be on the common scene in less than 3 years.

  8. dalexander
    September 28, 2012

    @Mr.Roques, With common language and open architectures being increasing introduced into the world of automation, it isn't hard to find or create the API (Application Program Interface) that will tie the physical equipment to the voice technologies. Having said that, the winner of the biggest piece of the pie will be the company that can interface with the most equipment already in play. And, set the standard for the interfacing protocols.

  9. dalexander
    September 28, 2012

    @Stochastic, the would seem like the response to t. alex's comment about working in a noisy environment. Does the solution you describe start with voice recognition or side band cancellation techniques. Can you write a bit more on this technology? When people get stressed, their voices tend to modulate up. Please write more on this. Thanks!

  10. dalexander
    September 28, 2012

    @Barbara, When the Manufacturing System software sends a command and the recipient performs the task, there is a confirmation input via a voice response from the picker. As soon as that confirmation response is received, the software updates the database record and time and date stamps the transaction in real time. So records are updated immediately with an ideal audit trail, including the picker's name and action type.

  11. stochastic excursion
    September 28, 2012

    Voice recognition is a vast field; the academic disciplines that go into it are deep and wide, and a big chunk of the advances made are by countless small research centers, both public and private.   The general approach as I understand it is taking the amplitude envelope of common vocalizations, along with the audio spectrum content of these sounds, and getting statistical signatures for the waveforms and peaks in the audio spectrum.  

    Fundamentally, speech waveforms are modeled using Markov statistics.  Like Gaussian statistics that predict the shape of a bell curve, Markov statistics can reduce the description of speech to a few parameters, which turn out to be quite unique for a given speaker.  

    Beyond this, each development group has its own recipe for how it refines the recogition algorithms.  I'm sure additional spectral correlation goes into it.  A good source for finding out where the research is at in this area is the SBIR award announcements (for instance,

  12. dalexander
    September 28, 2012

    @Stochastic, fascinating. I am going to research Markov to get on the learning curve. Thank you for the comprehensive response.

  13. jonsiders
    October 11, 2012

    There are many voice applications but to be truly successful in a high noise environment you have to have a speaker dependant engine.  Voxware, Inc. has developed a specific voice recognition engine that is trained on a specific person's voice.  It takes 20 minutes of training the system just one time, then keeps a profile of that persons speaking on their voice only wearable.  As soon as they log in by saying their name in the headset, their voice can be heard in airplane loud background noise.  So this allows voice applications in noisy freezers and manufacturing facilities, near forklifts, near conveyer belts, or any other high noise environments.  

    Voice with Voxware has been used in food distributors, retail clothing warehouses, restaurants, automotive parts and manufacturing, paint manufacturers, drug manufacturers, grocery stores, 3PL's, Publishing, and dairies.  We have the leading food distributors and auto parts companies in the nation currently using our technology in every distribution center they own.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.