Machine Vision in Logistics Applications

One of the earliest applications of machine vision was as a slug matrix code reader in distribution houses. Even before WalMart, companies like K-Mart had adopted these systems to route cased goods in warehouses. When a store placed an order for a complement of products, the computer system generated a pull list. A person physically took a case of goods from the warehouse shelves, placed the case on a conveyor and placed the routing label on the case. The case was then conveyed past an overhead scanner that would read the slug code, which designated to which truck it should be routed and essentially where on the truck it should be positioned so that when it arrived at the destination store, the entire truck did not have to be emptied to get to those cases ordered by that specific store.

The main driver for the slug code was that it offered the ability to incorporate more data while occupying less real estate than one-dimensional bar codes. One could argue that the slug code was the forerunner of the 2D symbolic codes that have emerged in recent years have been deployed in logistics related applications. One of the more popular codes used in logistics applications is the Maxicode pioneered by United Parcel Services. Maxicode looks like a honeycomb with a bull’s eye target in the middle. PDF-417, a stacked 1D code is also widely used for paper trail traceability in shipping and receiving applications.

Reliable reading of 2D symbols requires the ability to find, enhance and segment the 2D symbol as well as “read” it based on pattern recognition. In other words, all the functions one finds in a typical machine vision system. The “smart camera” implementation yields a low cost reader, whether hand held or fixed mounted.

In logistics applications one also finds 3D-based systems designed to provide carton shape information as well as dimensions. Dimensions and shape provide additional information to optimize the load of a truck.

To gain additional insights into the application of machine vision-based 2D symbol readers and 3D-based machine vision systems in logistics applications, companies offering such products were asked to contribute to this article by answering a series of questions. The following companies participated:

• Karl Gunnarsson – Vision Business Development Manager – SICK
• Mr. Gerhard Baer, Director Ident-Technology, VITRONIC

1. How would you describe your product line?
[Karl Gunnarsson – SICK] SICK is a leading provider of a wide range of innovative machine vision technology that is currently deployed in a variety of logistics applications. Machine vision solutions from SICK are divided into three categories: Vision Sensors, Smart Cameras (2D and 3D) and 3D Cameras.

Vision sensors, such as the SICK CVS line of Color Vision Sensors, which use 2D, are ideal for package sorting using color, verify label presence prior to in-line printing, identifying the presence of a mail in a tote/tray, and more. SICK offers other vision sensors for date code inspection, area gloss detection, and other applications.

2D Smart Cameras. The IVC-2D, when combined with a light source, processes objects using gray scale contrast to differentiate an object from its background. It offers precise measurement and accurate verification of size.

3D Vision Cameras and Smart Cameras. SICK offers several 3D technologies that are laser-based. Products include Ruler and Ranger Cameras and the IVC-3D Smart Camera. The Ruler Logistics detects the location of objects in sorter systems. With its innovative 3D laser technology, it is ideal for detecting very small objects without confusion from labels or low contrast situations. In addition, these products use triangulation technology, which is based upon the well-proven approach of measuring displacement of a laser line. Displacement is the function of object size.

[Gerhard Baer – VITRONIC] VITRONIC machine vision products are technically advanced products.  They are used as well in areas where solving complicated applications is required as in standardised areas such as identification of packages in the parcel distribution industry.  Often they are the result of applied research and development. One fine example is the development and production of VICAM ssi, the ultimate auto focus line scan camera with integrated LED illumination for image acquisition of multisided package surfaces.
VITRONIC supplies a state-of-the-art system for multi-sided package identification:  the identification system VIPAC reads hand-written and machine printed address information, identifies barcodes and 2-D codes, measures the volume, and enables manual corrections of incomplete address information to be made at video coding terminals.

2. What specifically differentiates the models you offer?
[Gerhard] VITRONIC products are differentiated in many ways across the product line.  Our products are designed and manufactured in-house.  VITRONIC as a company is ISO 9001 certified and in markets that require certification such as for volume or quality, the products meet the necessary standards and will be supplied in accordance to the local certifications.  With VITRONIC’s long-term experience and the thousands of systems sold we provide industry proved and tested technology.

[Karl] Machine vision cameras from SICK integrate imaging technology, lighting, display and evaluation hardware and software into one compact housing.  Our sensors and cameras improve product quality, production yield and reduce the costs of inspection. Our strengths include the breadth of our innovative solutions and ease of use. From introducing the world’s first 3D Smart Camera… to our super high speed Ruler E with Gigabit Ethernet Interface…to the powerful, yet easy-to-program CVS Vision Sensors, SICK can design a solution to meet our customers’ exact needs. 

3. Would you characterize machine vision-based systems used in logistics applications as 'hardware-specific' or 'software-specific' and why?
[Karl] Neither. Describing functionality as software-centric could imply that that the hardware plays no significant part in the solution. For SICK, our 3D solutions (IVC and Ranger\Ruler) are based upon a unique CMOS sensor with on-board RISK process. Other vision systems are designed around dedicated hardware. However hardware alone does not solve applications. The ramification is that we have invested heavily in novel ways of processing images generated by the hardware. With an in-depth understanding of the hardware, the task of writing effective software that will fully utilize the power of available hardware is possible.  The implication is, therefore, that the functionality is evenly distributed between hardware and software.

[Gerhard] This is like the 80/20 rule.  Machine vision describes hardware; however it is the software that makes the hardware meaningful.  Advanced algorithms for recognition and performing functions such as signaling sortation or collecting statistics or making decisions about quality, etc. are software-specific and absolutely required.

4. What are the critical parameters of a machine vision system that a logistics customer should understand? If you offer both 2D and 3D systems, review parameters of both.
[Gerhard] 2D applications are usually simple compared to 3D applications.  It is similar to the difference between a car with a gear stick and a car with automatic transmission.  The result is the same but one requires less human intervention in the process, and more invention in the design.)

Read rate, label size, object size, depth of field speed scan rate, the accuracy and the field of view are counted among 2D critical parameters. 3D parameters in logistics are length, width and height information of the relevant package. This 3D information is needed for both setting the cameras’ auto focus and for customers’ information of package volumes. These volume measurement systems provide certified results for invoicing etc. All these parameters have a direct impact to each machine vision application.

[Karl] 2D: With 2D, the critical parameter is minimum allowed contrast difference between object and background. A 2D system will work well when there is enough contrast difference between the object and carrier. Speed, resolution, interface, bar code type, etc. are all parameters that need to be understood.

3D: Height, width and length resolution, are important parameters that affect a 3D system. The length resolution is the least understood. Length resolution gives users the ability to accurately locate the object in the movement direction. This is important for a logistics application because it indicates where products are at all times. The SICK Ranger and Ruler Cameras have a very high frame rate, making them fast and accurate.

5. What are the skills required to program and operate a machine vision system in a logistics application?
[Karl] Machine Vision products from SICK are designed in such a way that no vision or in-depth understanding is needed to operate them. Our IVC Studio Software for the IVC-2D and 3D provides quick and easy access to more than 100 powerful image-processing tools. Once configured, the camera can work in stand-alone mode, with no PC required. CVS Vision Sensors offer an integrated color display and keypad that makes it easy to perform the teach-in procedure. After initial setup, they can be trained remotely via a PLC. The Ranger and Ruler Cameras use configuration and evaluation tools and require .Net, C++, VB development skills.

[Gerhard] Translating operational processes into automatic processes is very important.  This includes both operational issues about product flow, sortation, packing, picking and loading bays to business rules such as shipping rates and alternatives for cost reduction in shipping optimization. This is VITRONIC’s part. Running a machine vision solution at site doesn’t require special skills of our customers.  They should just understand sorter technology.  Programming at site is not required as all components such as cameras and dimensions are supplied pre-adjusted and pre-calibrated, including the necessary software engines.

6. How do you support your products – training, documentation, warranty, post installation service, software revisions? Are these free or is there a fee?
[Gerhard] VITRONIC machine vision products come with a standard warranty and optional maintenance agreements are available.  Each program is tailored to the requirements of the customer. The customers’ service staff will get operational and maintenance training including corresponding documentation. The training sessions will be conducted by our application engineers at the customer’s site. In addition VITRONIC provides a hotline service throughout the globe and offers preventive maintenance as well. This will include new software revision as well.
[Karl] SICK provides support for before and after field installation through an established network of field personnel. SICK can also provide product training, start-up assistance and service and repair. Rarely, does SICK does offer turnkey systems for complex machine vision applications. We work with end customers and integrators to create customized solutions. Product training is offered both on-site and in a classroom setting. The warranty is standard for the industry, and software revisions for the products are free.

7. Where do you see breakthroughs coming in the specific infrastructure technologies (hardware and software) that are the basis of the 2D or 3D-based machine vision systems used in logistics applications in the near future – next three years?
[Karl] Software will make products easier to use. The SICK-IVP CMOS vision technology will continue to provide distinct advantages for 3D applications. For 2D vision applications, I do not foresee revolutionary changes.

[Gerhard] One interesting area is increasing information density.  As 2D codes become more common, users will be empowered to add much more information on the labels.  It is the integration of machine vision into other disciplines that will offer breakthroughs such as vision solutions, having dimensioning and/or weighing integrated.  These additional parameters enable distribution companies to optimize logistic processes such as revenue recovery, automated invoicing and intelligent truck loading.

8. What specific performance improvements are anticipated driven by these forthcoming technological changes? How will they impact the use of these systems?
[Gerhard] The customers’ goals do not change very much over the time:  They always lead to “how can we optimize our processes in order to achieve higher throughput at higher degree of automation at a shorter return on investment?” Does it help to generate USP’s for me to my customers?  Does the ROI justify implementation of automated identification systems over labor costs or contractors?

[Karl] Performance improvements will mainly be in increasing resolution and sort accuracy.

9. What are the main markets using for these 2D or 3D-based machine vision systems for logistics applications of one type or another and are there market changes within those industries that are driving the adoption of machine vision systems?
[Karl] There are numerous possibilities for logistics applications. There are three main classes of conveying mechanisms for consideration: cross-belts, continuous belts and trays. The overall requirements are the same – move more products per hour with fewer mistakes.  3D provides better object position information. This enables companies to better control and sort information and products, saving time and money.

[Gerhard] The two most important areas are in the parcel and package shipping segment and distribution centers.  In the first case, accelerated fuel costs have recently initiated massive cost reduction programs.  Vision systems reduce cost due to higher degree of automation and increase productivity due to higher throughput.  In the warehouse or D.C. world, machine vision is integrated into “smart picking” processes, which are reducing manual pick methods and their costs dramatically.  Imagine the cost reduction from an average time to pick a product reduced from 12 minutes down to 25 seconds!

10. What impact do these changes have on the technical requirements (specifications) for the 2D- and 3D-based machine systems used in logistics applications? And how will these machine vision systems have to change to address these more demanding requirements?
[Gerhard] Machine Vision Systems in logistic applications offer system integrators the opportunity to invent new conveyor and sorter technology. They can be operated at faster speeds.  It also means that simulations of a hub layout will play an increasingly important role in design phase.  Due to advanced identification technology such as six-sided reading and 3D dimensioning systems the parcel inbound can be automated to a much higher degree. This results in lesser amount of staff. These six-sided identification systems become more and more common at the market place. This offers system integrators possibilities of improving the hubs’ overall dimensioning amongst other. 

[Karl] All conveying systems need to be able to handle varying types of packages at maximum speed. To enable this, the vision system must, no matter what size, color, background or package type, be able to provide package position and number of objects on a conveyor, tray or cross-belt. This is a possible, but challenging for black objects on a black background and for objects that are 1x1x1/4 inch running at 2 m/s, for example.

11. As a supplier of 2D and/or 3D-based machine vision systems for logistics applications what are some challenges you face in marketing machine vision systems?
[Karl] 3D technology for machine vision is still not well known or well understood.

[Gerhard] The main challenge of our marketing efforts is to create a wide understanding of the advantages when implementing machine vision versus standard technology such as scanners. This specifically means that machine vision offers higher degree of automation for multisided applications as the same images can be used for barcode reading, but also for 2D code reading, address reading, video coding, and last but not least for image storing of parcel surfaces for later claim issues. The systems also offer possibilities for automated billing and revenue recovery. 

12. What advice would you give to a company investigating the purchase of a machine vision system for logistics applications?
[Gerhard] Trust but verify.  Be certain that you compare or investigate multiple vendors before purchasing decisions.  Make sure that you have seen other machine vision solutions as references.  Analyze and document your current process and ask for additional ideas from potential vendors.  Vendors will have knowledge of what is working in other customer environments that can migrate to other similar own operations. Count on specialists with long-term experience in providing turnkey machine vision solutions in the logistics market.

[Karl] Communicate your requirements early and clearly. Requirements should be transformed into a solution where those requirements, performance and solution price are matched. Finding penny size packages on a 2 m/s conveyer is possible, but a bit too expensive for most customers. And, 3D vision if applied correctly can provide some distinct advantages compared to 2D.