iTECH: Virtual Dispatcher May Be Virtually Ready

By Dan Leone, Staff Reporter

This article appears in the February/March issue of iTECH, published in the Feb. 15 print edition of Transport Topics. Click here to subscribe today.

In the trucking dispatch room of the future, the only sound is of cooling fans whirring inside computer casings.  Gone are the cubicles that once honeycombed the floor, as are the chattering dispatchers who once staffed them. In their place, rows of powerful computers — imbued with dynamic, problem-solving algorithms that simulate the thought process of their human predecessors — silently schedule deliveries, manage drivers on the road, and seamlessly incorporate late-breaking directives from shippers into the fleet’s workflow.

A lone human being keeps watch, monitoring his computer screen for the occasional emergency — breakdowns, accidents, massive traffic jams — that his team of “virtual” dispatchers cannot resolve on their own.

This is the ultimate dispatch center envisioned by purveyors of truckload software. But how realistic is this vision?

Ironically, the enabling technology for the virtual dispatcher is older than deregulation, the head technology executive at one software house told iTech.

“I think the technology is ready for a true virtual dispatcher,” said David Mook, chief information officer and chief operating officer of TMW Systems Inc., Beachwood, Ohio. “I think it’s here today, that it’s been here.”

In fact, Mook said, “the core algorithms” needed to power a virtual dispatcher “were invented in the ’70s” — years before the Motor Carrier Act of 1980 disassembled federal control of interstate trucking rights.

Algorithms are essentially problem-solving flowcharts that software uses to approximate the human thought process. They also are the germinating seeds of the virtual dispatch of tomorrow.

The virtual dispatcher of today, such as it exists, is in an embryonic stage that software providers call “load-matching optimization” or sometimes “dispatch optimization.” Many of the largest trucking companies in the country already use such technology, Mook said.

Load-matching software is designed to pair available drivers with available loads in the way that is most profitable for the carriers. Drivers’ location, remaining hours, desired home time and other factors are all taken into account, just as they would be by a flesh-and-blood dispatcher.

Load matching is available today from TMW and Manhattan Associates, Atlanta. These two vendors are the preeminent providers to trucking, said TMW Chief Executive Officer David Wangler.

The founder of McLeod Software, TMW’s biggest competitor in the niche world of truckload software, said that the main impediment to fully automating the dispatch center and turning the stage over to virtual dispatchers is something that plagues man and machine equally: incomplete data.

“For the computer to make the ideal recommendation, you have to have all the information in the system” before a driver is even dispatched, said Tom McLeod in Birmingham, Ala.

“You need every factor: When does the driver need to be home? When will the driver have to take a break? What are the loading and unloading time windows? What’s the ideal routing?”

McLeod Software does not have an optimization engine of its own — it relies on an integration with optimization packages from Manhattan Associates — but as the second-largest provider of off-the-shelf dispatch software in the country, McLeod has some experience integrating the subsystems needed to feed a virtual dispatcher all the information it needs to make this kind of business decision.

McLeod pointed to several key components:

n Two-way machine-to-machine communication — This is a big part of the picture. Shippers’ computers must be able to send freight data automatically to carriers’ offices via electronic data interchange — or some other M2M language — so that human beings at the shipper’s office don’t have to call or fax human beings at the carrier’s office.

n Dispatch software — Keeping track of active and idle loads, and active and idle drivers, on a white board or legal pad is impossible for a computer. Bringing dispatch software into the picture provides a repository for this information that can be accessed and evaluated by both human and virtual dispatchers. The right dispatch software also must have M2M communications capabilities, so that a virtual dispatcher can broadcast its assignments to drivers.

n In-cab mobile communications — Not just any system will do. To facilitate automation, mobile communications must be able to integrate with the carrier’s dispatch software. With such a setup, the virtual dispatcher can eliminate the need for a human dispatcher to make a phone call to a trucker on the road.

n Global Positioning System tracking — One of the first questions any dispatcher, virtual or human, must resolve before ordering a pickup is, “Which driver is closest to this freight?” It is impossible to know the answer without GPS.

n Optimization engine — This essentially is the virtual dispatcher itself — a computer program built upon problem-solving algorithms that can identify the most profitable hauls and factor in many decisions, given the constraints McLeod mentioned.

Mook at TMW added electronic onboard recorders to that list. Knowing how many hours a driver has left is essential to determining whether that driver is the right match for a certain load.

Mook put it like this: “If the driver calls you and says, ‘I only have four hours left,’ you don’t give him a load. If his mobile comm unit says he has eight, you can make him run a load.”

Mook told iTech that one U.S. fleet, which he declined to identify, has strung all of these tools together and designed an experiment to test whether assignments generated by software that creates the optimal dispatch actually result in a more efficient and profitable trucking operation than human-assigned loads.

The fleet, a truckload carrier, is conducting a trial of “one thousand trucks in one thousand lanes,” Mook said. The trial hinges on Match Advice load-planning software, which TMW acquired in 2007 along with Integrated Decision Support Corp. Among other tasks, the software products calculate driver-to-load matches and chart routes that offer optimal fuel use and purchases, as well as turn-by-turn navigation.

More precisely, the trial hinges on drivers complying every time with the automatically tendered recommendations from Match Advice, Mook said.

That means no calling home to the dispatch office to beg for a different load and no cherry-picking from the computer’s list of recommendations. Drivers must take the load that the computer determines is the most profitable.

The trial was in progress at the time of this writing.

The experiment Mook described provides a subtext for the current state of machine-human relations at trucking companies today.

Namely, that they are still a bit rocky.

Mook realistically estimates carriers that use load matching software — the proto-virtual dispatcher — let the software pick “only about half of their loads. The other half of the time, someone is saying that the computer is wrong.”

Even if fleets instituted full compliance with recommendations from their dispatch programs, no algorithm on the market can cope with unplanned incidents, such as accidents and breakdowns.

These events are known as “exceptions,” and at present, “we don’t have a very good way to handle the exceptions,” said McLeod. And there are “a tremendous amount” of exceptions, he added.

The best that automated dispatch systems available today can do is monitor for these exceptions and alert a human being when they occur.

With GPS tracking and mobile communication, a truck can inform the home office, without human intervention, that it has made an unscheduled stop — but only a human being can do something about the unscheduled stop, McLeod said.

Given the current state of the art, he said, a single dispatcher can manage anywhere from 40 to 80 vehicles — a far cry from a dispatch room staffed by only a single human being, but an improvement over the early 1990s, when a lack of mature communications systems, dispatch software and integrated algorithmic decision-makers meant that one dispatcher could handle no more than 25 trucks.