When triggering the systems that apply the throttle, apply the brakes, and follow highway curves, highly autonomous vehicles rely on a vast array of data.
Whereas truck drivers behind the wheel of conventional vehicles rely on their senses, the autonomous vehicle's technology interprets its surroundings using cameras, radar, and LiDAR lasers. Radar waves monitor the distance between the truck and surrounding vehicles, while LiDAR generates a precise three-dimensional map of the environment. Observing lane markings and contours are cameras.
Artificial intelligence (AI) brings everything together, processes the data, and determines the appropriate response.
While the majority of AI-related conversations in the transportation industry center on autonomous vehicles, the technology has already made inroads in other industry applications. In the process, the guard at the yard gate, office workers, and terminal administrators receive a futuristic makeover.
Decisions based on data are assisting safety managers in reducing accidents by highlighting hazardous driving behavior and highlighting positive actions. Moreover, efficient dispatch systems optimize routes, boost earnings, and increase the frequency with which drivers return home, thereby enhancing driver retention.
Jessica Kim, director of marketing for Pitstop, a provider of predictive fleet maintenance software, explains that artificial intelligence does not replace humans or eliminate jobs. "Artificial intelligence gives people superpowers that increase their productivity."
When making deliveries or gathering up freight, drivers must traditionally check in with a gate guard. Or, they may be required to enter a shipping office where a worker processes documents and directs them to the loading bay.
Patrice Boies, vice president of business development and partnerships at Nuvoola AI, asserts that AI-based decisions can make the completion of such duties cheaper, faster, and more accurate.
Instead of relying on human eyes, cameras and software can identify vertical or horizontal characters, license plate numbers, trailer serial numbers, and equipment names and numbers, while software organizes this information into records.
The company is conducting tests with a smart kiosk that processes visitors by recognizing biometrics and accents, as well as processing documents. Even in real time, its 'natural language comprehension' translates languages.
Will the security shack and shipping office personnel become redundant? "Our goal is to improve efficiency and productivity, not necessarily to replace humans," Boies explains.
He claims that the kiosk provides a return on investment. The average hourly wage for a security officer or employee could be around $25, rising to approximately $40 per hour when benefits are factored in. "Depending on whether the contract is for 24, 36, or 48 months, operating the kiosk could cost as little as $1 or $2 per hour," he says.
He adds that the officer would then be free to support facility security and conduct walkarounds.
Such systems can even examine a truck's cargo in greater detail. For instance, the data scientists at Nuvoola are investigating methods for weighing log cargo, measuring the distance between them, and examining their rings to determine age and product density.
"There are no limits to technology," asserts Boies.
Pitstop employs predictive maintenance to assist managers in determining which automobiles to herd into service facilities. Pitstop's chief technology officer, Vedant Khattar, explains that the first data source is unprocessed and includes telematics, fault codes, battery and cranking voltage, and lubricant temperature. "It comes from any sensor that we can access," he explains.
Service history or field data includes details such as the maintenance conducted by the mechanic and the parts that were replaced. Then, driver vehicle inspection data and sensor readings are compared.
"[We] have developed algorithms for engines, brakes, batteries, and fault codes," says Khattar.
Then, intelligent maintenance schedules can provide alerts in addition to actionable and understandable reports. "Pitstop is the bridge point that not only pulls in data from equipment manufacturers and technical service partners, but also translates it into something the end user can understand and react to," says Kim.
In the end, AI's true power emerges when data are combined to forecast a need.
Using geo-tagged data, such as weather, traffic, and even forest fires, shippers and receivers could determine why a motorist is late.
Human employees make decisions using comparable data, but the process is time-consuming.
When reviewing raw data, fleet managers may also make usage-based decisions, such as changing the oil every 'x' miles. Those with telematics systems can rely on automated triggers to be notified when a vehicle reaches a predetermined number of miles or hours. But AI contributes additional data to determine precisely which vehicles require attention.
For example, Pitstop software labels alerts as critical, major, or minor. A critical alert means the vehicle must undergo maintenance immediately, a major alert means it can wait until the next scheduled preventative maintenance (PM), and a minor alert can wait until the next PM.
The more information that these systems accumulate, the more precise their decisions become.
"Access to data is required to train models," says Khattar. "After six months, we anticipate an accuracy percentile of 95%, and the AI flywheel is continuously improving with feedback."
Kim of Pitstop emphasizes that individuals will still need to be involved.
"We are your analysts," she proclaims. "Artificial intelligence is only as effective as the people who use it and act on its findings."