Drilling advances

As artificial intelligence (AI) makes its dispassionate presence felt within the drilling ecosystem, might it also be used to lower one of the industry’s more sobering statistics: driving-related fatalities?

No, we’re not talking here about the improbability, anytime soon, that driver-less vehicles will magically transport mud engineers, frac crews and the like, to and from locations, but rather the prospects of AI-enabled technology being harnessed to ensure that those actively engaged in the motoring process avoid enroute siestas. To point, one of several articles encapsulated in an extensive Wall Street Journalreport on AI, on April 2, detailed how AI developers are working with vehicle manufacturers to integrate devices that can identify when drivers are becoming drowsy and provide an immediate wake-up call. On a far-less serious note, a companion article examined an AI development that a contractor or service company’s sales organization conceivably could use as the scapegoat to mollify disgruntled clients (more on that later).

Meanwhile, quoting U.S. Department of Labor (DOL) statistics, the National Institute for Occupational Safety and Health (NIOSH) reported in March 2018 that more than 40% of oilfield fatalities can be traced to motor vehicle crashes. According to NIOSH, most of the blame for vehicular deaths can be laid on driver fatigue, the result of insufficient sleep, long work shifts and often long distances between wellsites and domiciles. A few years ago, an Oklahoma-based drilling fluids engineer told me that driving 300 mi was a routine day, which was on top of the time spent checking mud on the rigs within his jurisdiction. Aggravating the napping driver syndrome are badly damaged rural roads that were ill-designed for the heavy traffic typifying the shale phenomenon.

During an earlier life, countless ride-alongs to rigs throughout West Texas, Siberia and elsewhere, instilled an appreciation for the rigors that these road warriors endure. I especially remember one posterior-numbing endurance test, sitting in the back of a jeep traveling from Cairo to rigs in Egypt’s Western Desert, where only an oasis and occasional petro station broke up the eye-drooping monotony of gazing for hours at endless sand vistas.

While companies have long combined training, journey management systems and in-vehicle monitoring to record excessive speed, fast braking and other untoward driver behavior, data show fatigue continues to be a major issue.

AI to the rescue? The aforementioned WSJarticle examined how automobile manufacturers and AI companies are combining their expertise to design systems that use voice analysis and facial recognition to assess “the alertness and emotional state of the driver.” Last year, Toyota unveiled its Concept-i prototype that uses an infrared camera on the steering column, a pair of 3D sensors on the instrument panel and an onboard speech-recognition and conversation system to read facial expressions and voice tones. If the motorist exhibits signs of drowsiness, the integrated system reacts and immediately engages the driver in conversation, as one-sided as it may be.

In what sounds like a Ray Bradbury fantasy, the article went on to suggest that over time, the conversation system will be refined to the point that it knows precisely what topics are most likely to engage the driver. I doubt that those topics will extend to quizzing the droopy driver on the rheological readings recorded in the latest mud check, or strike up a hearty debate on which PDC cutter configuration is best suited for the next interval.

However, according to the article, AI developers, Affectiva and Nuance Communications Inc., said last September that they were teaming up to add “emotional intelligence” into the latter’s Dragon Drive conversational automotive assistant, which in its current form, is already installed in more than 200 million cars of various brands. With the refinement, the proprietary algorithms would use deep learning, computer vision and speech technology to identify emotions and indicators of drowsiness and respond accordingly.

The algorithm did it. New-generation algorithms, likewise, could be sacrificed in the name of preserving quality customer relations, another article suggests. Under the headline, “Beware Algorithms That Could Collude to Unfairly Raise Prices,” the piece described how companies have turned to algorithms to help set prices for all manner of goods. All well and good, until Italian researchers recently conducted a series of computer simulations, which revealed that pricing algorithms learned, and through trial and error and without communicating directly, colluded to raise prices above competitive levels. The naughty algorithms were programmed to maximize profits, with how to do so left to their own devices.

With contractors and service/supply companies trying to recoup some of the steep discounts extended at the height of the bust, this opens an opportunity to head off an uncomfortable meeting with a client staring at an especially stout invoice. Say for example, you gamble and nearly double the daily rental on a centrifuge that basically had been billed at cost. The customer could grudgingly accept it as a cost of doing business and let it pass. Conversely, if said client has a widely different take, you could simply respond with “the algorithm did it.” Sure, your bottom line may not be as healthy, but at least you won’t be the fall guy.

Jocularity aside, the research also provides a cautionary tale for companies looking to give AI a louder voice in their operations. “The question, as we are delegating key managerial tasks to machines, is what shall we expect?” Emilio Calvano, professor of economics at the University of Bologna and one of the investigators, was quoted as saying. WO

About the Authors

Jim Redden

Contributing Editor

Jim Redden is a Houston-based consultant and a journalism graduate of Marshall University, has more than 40 years of experience as a writer, editor and corporate communicator, primarily on the upstream oil and gas industry.