PROBLEM:



Every year, the United States Postal Service (USPS) handles over 150 billion pieces of mail (almost 5,000 pieces every second). It handles more mail than the rest of the world's postal systems combined, posing a king-size problem in customer service. To deal better with this immense problem and to cut costs, the USPS is turning to technology and automation, specifically, optical character readers (OCRs) and electronic bar codes. In fact, by the year 1995, the USPS wants to have bar codes on 95% of all mail. Other than mass mailers, who will apply the bar codes themselves, the USPS will have to read the address and then apply bar codes at postal facilities.

The magnitude of the problem becomes clear when considering just one aspect: the handling of IPPs (irregular parcels and pieces). These include mailing tubes, film mailers, and boxes of all shapes and sizes. A postal worker currently sorts 200 such parcels per hour, picking pieces off a large table and putting them into stacks. To automate the sorting process, the USPS is considering robotic handling systems coupled with machine vision that would identify individual packages on the conveyer, calculate physical parameters (e.g. center of gravity and position), find the label, and read the ZIP code. In addition to locating ZIP codes, the USPS eventually hopes to perform a contextual analysis of the city-state part of the address; the idea is first to read the city, state, and ZIP, then cross-check all three from a huge, nationwide database.

Unfortunately, optical code readers can only read about 40% of all mail, and current experimental recognition systems are able to decipher only about 30% of the remaining 60%, Furthermore, the scanning technology will cost the Postal Service at least $24 million, according to officials from Intermec of Everett, WA.

There is another aspect to the problem of reorganization: accessibility. While certain competitors offer "mailboxes" where small, thin packages can be dropped off at any time, one can mail packages (of any size) through the Postal Service only during post office hours. Furthermore, long lines often increase the time wasted waiting for a free clerk. Thus, it can easily take more than half-an-hour just to mail one package. As a result, many people are fed up with the inefficiency of the Postal Service. Perhaps most indicative of the USPS's problems is that, in 1988, its share of the package market had dropped to 8%. If the problem is not dealt with adequately, that share may drop even lower.



SOLUTION:



In approaching this complex situation, we did not attempt to reorganize the entire postal system. Rather, we chose to focus on the issue of IPPs (packages, parcels, etc.) and came up with a solution that facilitates the implementation of automation and increases accessibility at the same time. To make mailing packages more convenient, we examined the way banks have used automatic teller machines and conceived a similar idea for the Post Office: an Automated Postal Machine (APM). This device would enable people to mail packages at their own convenience without the assistance of a postal clerk. Whether at 12 noon or 12 midnight, one could send packages without waiting in line, as well as at night when the post office is closed. Furthermore, we realized that the APM could also help improve the automation of IPPs if a bar code label were placed on a package as it was accepted by the machine. With the implementation of a network database, each package could be given an individual code by which it could be tracked on its route from originating post office to destination. The IPPs avoid spending money on implementing expensive optical character readers and similar machines because people enter the information into the machine directly.

We have designed the APM to accept most parcel post mailings, which are defined by the USPS as all fourth-class mail weighing one pound (0.45 kilogram) or more that do not qualify for mailing at special rates. The current regulations limit the maximum size and weight of a package, the APM can handle parcels no larger than 2.5 feet (76.2 cm) in width and 2 feet (61 cm) in height. Nevertheless, we believe this will accommodate most standard-size parcels.



THE APM:



In order to design a relatively accessible automated machine, we once again used ATMs as a model. In fact, a number of people gave us funny looks as we took some measurements of the local bank machine; they probably thought we were planning a robbery. We found that the machines are generally 5 feet (150 cm) tall and 2.5 feet (75 cm) wide, with the screen and keypad situated about 4 feet (120 cm) off the ground. While we understood this design had been chosen after careful study to be the most convenient, we also realized that the ATMs were not designed for depositing parcels. As a result, we decided that a slightly wider frame, 3-feet (90 cm), would be required. Furthermore, because the opening and package bay are bigger than any ATM deposit space, we were concerned that small children could climb into the package bay while their parent was busy using the machine, find the door, close and latch it behind them, thus trapping them inside. Therefore, we tried to make the bay door as high off the ground as possible without pushing up the screen and keypad. In addition, we agreed that the bay door would not close should the electronic scale sense a weight above 25 lbs (11.3 kg).

Once the bay door has been safely closed and the package's weight verified, the back of the package bay opens, and a mechanical arm about 2 inches (5 cm) in height would emerge from one side, reach over to the other, and then sweep along the scale from front to back (like the pin-clearing arm at a bowling alley). Thus, any size package, whether only a few inches thick or a few feet tall, is swept out of the APM and onto a conveyor that brings the package back into a storage area from which it can be sorted.



THE CONTROLS:



The centerpiece of the APM's input system is a touch screen coupled with a ten-digit keypad. These components will enable novices to use the system comfortably.

A typical session will commence when the user touches the screen, changing the screen from its welcome/screensaver mode to the account-card verification screen. The customer will then have to provide a debit card for verification. When completed, the user will be asked which transaction to perform: sending a parcel or checking on the location and estimated time of arrival of a package already en route. If the customer chooses to begin mailing procedures, they will be prompted to open the bay door and to place the parcel on the scale. A new screen will ask the customer to provide information about the package, including the five- or nine-digit ZIP code, the contents, the declared value, and an optional COD amount. In addition, the sender will be asked to specify the mailing class desired (e.g priority mail). The customer can press the calculate function that will show the price for delivery.

When the sender has decided how to send the package, they will be asked if they would like to print a complete label. An image of the label would then be displayed on the screen, and the sender will be asked if all information is correct. After the customer has placed the label on the package and closed the bay door, they will be asked if they want another transaction. Their card will be ejected with a receipt showing all the data entered, including the parcel's identification number, and the APM will return to the opening screen.

If the customer has any problems, a 24-hour 1-800 number will be available for customer assistance.



THE LABEL:

When the customer has finally decided how to send the package, a label with an adhesive side is dispensed. In designing the APM package label, we had to consider two important factors: the setup of the package code and the ability of automated sorting machines to locate the label on a package. For the latter issue, we have proposed two solutions. One involves the placement of a special chemical or attracting tag (which could be detected by bar code scanners) on the label. The other suggestion, similar to the first, would use special fluorescent ink that could be detected by infrared scanners. This solution avoids spending millions on optical and mechanical equipment.



THE DATABASE:



In designing the package identification code, we had to consider the larger issue of how the network database would be established for tracking IPPs. We decided that instead of making one central, nationwide database that would handle all packages, local data centers for each two- or three-digit base ZIP code would be more practical. Thus, the exchange of information on a parcel would take place between the parcel's originating center and all the centers along the way to, and including, the destination center. For example, if one were to send a package from Boston to Texas, the Boston database center would receive information from New York City, Dallas, and the suburb of Dallas, the package's destination. At each point where the IPP is scanned, information is sent back to the Boston database. As a result, we formulated the package codes to include the ZIP code of both the originating post office and the destination post office.

For example, let us consider a package being sent from Newton Centre, Massachusetts, to Oakland, California, specifically, from 02159 to 94618-3234. Thus, the package ID number might be:

02159 0749 94618 3234

The A0749' is the random package number generated by the local database. In a sense, each package has two identification numbers: one that tells the origin and identity of the package and the other that describes its destination.

The label dispensed by the APM would include this information at the bottom in both numeric and bar code form (see illustration). A numeric code designating the IPP's particular mail class would be included on the label, as would a description of the contents. A two-digit code should be enough to cover these two topics together (i.e. the first number would designate mail class and the second contents). Furthermore, we are also considering installing touch screen keyboards into APMs that would allow senders to key in the name and address of the receiver, in addition to the ZIP code. This information could be included with other package data stored in the data base under each identification number. Thus, someone sending two separate packages to Walla Walla, Washington, could differentiate between the two more easily when checking on the status of the packages.



SAFETY:



A major concem in designing the APM was how to prevent the misuse of the machine. Because the APM is a public automated machine, ensuring its security is a formidable challenge. However, we were not as concerned with major terrorist bombings as with pranks being played using the APM. Young kids may see the APM as a good opportunity to mail a few big water balloons or mice, for example. A video camera would be installed with each machine to monitor and record the actions of the customer, much like an ATM's camera. Since the APM requires the insertion of some sort of debit card to open the bay door, the machine restricts use to those who have a sound credit account. Thus, any perpetrator is linked to the credit card they use to access the machine; in a sense, the APM is actually safer than a mailbox. Hopefully, these factors would be enough to deter any abuse of the APM.



NEGATIVE EFFECTS:



With any new technology, there are always advances and drawbacks. For the most part, we are pretty confident of the APM's effectiveness. At first, we were concerned about the implementation of the APM system, for we realized it would require a major commitment by the USPS. However, a few APMs could first be opened in various trial sites around the country in order to test the worthiness of the proposed system. Once the number of APM-equipped post offices increased, the network database could be established and the APM system could be fully realized.

Perhaps our biggest worry is the mechanical stability and durability of the APM. After all, it is still a machine, and quite a complex one at that. We are concerned about how it will stand up to wear and tear from customers dropping heavy packages on the scale, dirtying the scale surface, jamming the label dispensing slot, etc. Yet, if one looks at photocopying machines, for example, one can see that careful refinements and advances have the ability to make an incredibly intricate device run quite smoothly. An APM would certainly be not be as complex mechanically as a photocopying machine, so we hope that any technical difficulties will be overcome.



CONCLUSION:



We believe that our proposed Automatic Postal Machine and network database put the USPS on the right track in dealing with reorganization of its IPP-handling system. The Postal Service can thus avoid spending millions of dollars on machines that try to decipher handwritten addresses while simultaneously increasing its customer service. Furthermore, the introduction of bar code labels and package codes will help the USPS catch up with some of the services offered by its competitors. Having missed the opportunity to be a leader in the area of electronic mail, the USPS now has the chance to reclaim its place at the forefront of the package-shipping industry. The APM system sends the Post Office confidently into the technological age and opens it up to many exciting new possibilities. u