Connected services for fleets in ‘Auto Pilot’ mode

TU-Editor

---

30th July 2018

Susan Kuchinskas looks at opportunities to serve the autonomous fleet of the future.

How will fleet telematics services change with autonomy? Some services will obviously disappear: good-bye, driver coaching and sayonara, speed and hard braking alerts. With no driver to handle fueling, cleaning and maintenance, on the other hand fleet management applications will become critical.

In today’s taxis, if a passenger leaves trash behind, the driver will curse and clean it up. In an autonomous taxi, the trash is there to greet the next passenger. If a transport network company (TNC) vehicle gets a flat tire, the driver is responsible for fixing it. In an autonomous TNC fleet, the fleet operator needs to not only get an immediate alert but also be able to easily arrange for repair. “If I have a fleet of self- driving cars in a city, no matter how good my software or technology is, there are important things that need to happen,” says Sean Behr, president of Stratim. “These are basic and obvious but mission-critical.” Stratim, recently acquired by KAR Auction Services, provides a platform for highly distributed urban fleets, including on-demand car sharing and riding services, urban commute providers and autonomous vehicles.

Behr believes that fleet maintenance services will become more valuable for autonomous fleets, because of the cost of AVs. “Assets will be more expensive in the future and there’s a need for those vehicles to operate on an almost round-the-clock basis. Every minute that one of these autonomous vehicles is sitting around idle is a minute of lost money,” he says.

Understanding the total cost of ownership or operation will be more important for operators of autonomous fleets. The wealth of real-time data from connected AVs will make it easier for fleet owners to understand how much it costs to maintain a vehicle, how many trips it needs to make or how many passengers it needs to carry to turn a profit. “This will be critical metric of success in the autonomous world,” Behr says.

DSRC and autonomous fleets

Vehicle-to-vehicle communications is arguably one of the simplest and most productive connected-car services that’s also important for autonomous operation. While some manufacturers are independently moving ahead with DSRC, in the United States, NHTSA has been slow to decide on whether to mandate the technology.

The NAFA Fleet Management Association has urged NHTSA to move forward with rule making on DSRC, according to Pat O’Connor, a partner in Kent & O’Connor, a lobbying firm that represents NAFA. “We support, from the fleet perspective, a mandate for DSRC on passenger vehicles. There are a lot of benefit to fleets in reducing accidents and fatalities,” O’Connor says. There are also productivity gains to be had as DSCR becomes embedded in more and more vehicles, he believes.

In addition to supporting a mandate for vehicles under 10,000lbs (4.5 tons), NAFA also has advocated for extending the mandate to cover medium duty trucks and commercial vehicles.

The organization is aware that such mandates add to the manufacturing cost of the vehicle, costs that would be passed on to buyers. “From a fleet manager’s perspective, we have to be able to justify increases in our acquisition budget that result from increased costs owing to regulation but, as an organization, we’ve decided that DSRC makes a lot of sense,” O’Connor says.

NAFA is concerned with ownership and access to the data generated by autonomous vehicles. O’Connor says its position is that the data created belongs to the vehicle owner or lessee, as opposed to the automaker. “We share the concerns of the auto aftermarket repair sector that if all data were to be under the control of the automaker, that would not be a good situation for fleets,” O’Connor says.

Getting along in a mixed environment

Human beings, obviously, don’t drive algorithmically and can act erratically. As autonomous vehicles begin to mix with conventional cars, following the lanes and obeying traffic signals won’t be enough. In addition to being able to detect what’s around them, AVs need to understand how humans drive in different conditions, seasons, locations, times of day, regions, and more. Zendrive’s proposed solution is to train autonomous systems on human driving patterns.

“We’re unpredictable creatures. AVs need all the data they can gather to best predict our behaviors,” says Noah Budnick, data practice and policy director for Zendrive, a company that aggregates and analyses driving data derived from mobile phones. Budnick says data on driving behavior also needs to be localized, so that AVs can fit into local driving patterns. For example, the Pittsburgh Left custom allows a driver making a yielding left turn at a traffic light to go before oncoming vehicles. If the AV didn’t “know” about that custom, it might start through the light when it turned green.

Time of day is another important variable to take into account. For example, at a certain intersection during evening rush hour, people may be more likely to either stop short as the light turns red or cruise through. “That information can be used to train autonomous algorithms so that, if they’re at that time and place, they can be more alert.”

Zendrive is working with carmakers, Tier 1s and providers of fleet services to either provide an API for continuous data access or to provide one large data dump. It can pull data from specific geographic areas or time periods, allowing them to train algorithms around traffic flow and travel time.

Another potential services Zendrive could provide is independent safety scoring for autonomous vehicles or systems. “The hypothesis that autonomous vehicle will be as safe or safer than human drivers. But how do you know that? We specialize in measuring human driver behavior and you could use the same system to measure autonomous vehicle driving,” Budnick says.

As federal regulators set safety standards for AVs, Zendrive proposes being the independent third party that evaluates their driving data. It could not only compare various autonomous trips, since each trip will be different, it could also score AVs against human drivers. Will autonomous fleets become more than pricey experiments? Certainly but that depends on a new generation of services tuned to AVs.