Bryan Salesky, CEO of Argo AI, wrote that his company is currently designing self-driving cars to only operate within “specific areas of a city” and the autonomous vehicle system won’t be created for individual ownership.
“The system currently under development is not intended to be applied to vehicles that will be purchased and owned by individuals or to travel outside of their operational areas,” wrote Salesky.
Basically, the head of one of the largest autonomous vehicle (AV) companies in the world is saying that the dream of owning a robot car that will personally chauffeur passengers to any destination they want is unrealistic.
Instead, Argo AI is envisioning a system where Ford, one of their partners, owns a fleet of AVs that will operate in a confined area in certain cities, with the goal of expanding within each metro area and to other cities. Argo AI spokesperson Alan Hall says Ford's initial roll-out of its driverless fleet will be in Miami and Washington, D.C. He says Pittsburgh could be apart of the network eventually, but is not currently part of the initial roll-out, which is targeted for 2021. The driverless fleet will operate more like a taxi service, but only in a confined area.
Argo AI's messaging is different compared to the other local AV companies, like Uber, which marketed the introduction of AV testing in Pittsburgh in 2016. At the time, Uber envisioned a completely driverless future within 10 years, one where its fleet of AVs would be able to pick up passengers and ferry them anywhere they desired. The ride-hailing giant, which tests AVs in Pittsburgh, has since dampened those expectations.
“There's too much to perfect to enable a 100 percent autonomous future,” said Uber in a statement to Pittsburgh City Paper in January, “and we really do need the skilled expertise of drivers to handle the more challenging scenarios that are not ideal use cases for AVs.”
Argo AI is also hoping to combat the perception that driverless technology will initially be able to accomplish everything a human driver can do. Salesky wrote that the company is developing vehicles to be SAE Level 4, which means the car can drive itself in most environments, but with exceptions in certain types of weather and peculiar places. Hall says Argo AI vehicles will only operate on certain streets, at certain lower speeds, and during weather that is amenable to the AV technology. This means riders will be able to hail Argo AI AVs, but only within designated parts of cities, and possible not during some inclement weather.
Salesky wrote that Argo AI vehicles will require “further innovation” if they are to navigate in conditions like heavy rain or snow. He wrote that the company is working with Carnegie Mellon University to tackle those issues. Argo AI currently partners with Ford and will soon also partner with Volkswagen.
The CEO also detailed the AV company will be taking a “street-by-street, block-by-block” approach to developing the technology. Salesky wrote in 2017 that driverless cars wont be common on city streets anytime soon.
"At the end of the day, driving is local,” wrote Salesky. “Everyone has different expectations as to how someone should drive in a city. While our system cannot realistically accommodate every preference, we know that a self-driving car with only one set of driving characteristics will not fit every city.”
AVs use laser sensing technology, known as LIDAR, that is placed on top of the vehicles to map out city streets and surrounding environs. Hall says Argo AI is compiling that information so it knows where every stop sign and traffic signal is and is supposed to be, even if a stop sign falls over.
The AVs' cameras are then used to catch the movement of what Hall calls "dynamic objects," like pedestrians, cyclists, cars and other moving objects that are not static on the map. Hall says this is the real challenge for AVs and is the main reason Argo AI is trying to change people's expectations of what AVs will be when they are initially introduced.
"The hardest thing is predicting what the dynamic objects are going to do," says Hall.
So far, several AV companies have had trouble telling the difference between trees, pedestrians, and identifying other moving objects.
Some industry insiders believe that all of America’s streets must be mapped first using the technology, and then all that data must be put through simulation software so the computers that drive the cars can learn the likely millions of different scenarios that human drivers encounter over a lifetime.
A spokesperson at Aurora, another Pittsburgh AV company, told CP in January that its autonomous technology is put through thousands of software simulations daily. Aurora has been one of Pittsburgh’s more conservative AV companies in terms of what driverless cars are expected to accomplish in the near future.
And Salesky wrote that Argo AI wants to clarify the company's intentions and it is not looking to rush the introduction of AVs.
"While Argo intends to play a prominent role in the development of self-driving technology, it is worth repeating: this is not a race,” wrote Salesky. “Our street-by-street, block-by-block approach ensures that we will engage communities every step of the way.”
Editor's note: Initial reporting for this story did not include input from Argo AI spokesperson Alan Hall. His input has since been added.
Article also initially stated that human drivers would have to intervene with SAE Level 4 AVs, but Argo AI says this will not be the case because their AVs will only operate in a confined system that will not require human driver intervention.