Originally Published in Medium, Feb 14 , and updated on June 19th.
There are so many different opinions about autonomous vehicles (AVs) and when they will be available. Most of these are based on a lack of understanding. After three years of research and three books, my most recent being the second edition of Autonomous Vehicles: Opportunities, Strategies, and Disruptions, I am now confident of what is going to happen.
Autonomous vehicles will be available to passengers starting next year, with some preliminary rides available later this year. As I’ll explain, this will be available as autonomous ride services, not as individually-owned autonomous cars.
First some basics. There are three categories of autonomous driving: semi-autonomous driving, sufficiently-autonomous driving, and fully-autonomous driving. This correlate to the SAE definitions you may have heard about, but are a little simpler.
Semi-autonomous driving is advanced driver assistance systems (ADAS) that include adaptive speed control, lane centering, and automatic braking. This enables a car to drive autonomously on highways and other divided roads, but they can’t negotiate turns and corners. Many high-end models have semi-autonomous capabilities, but Tesla is the best.
Sufficiently-autonomous driving is the most important category right now. It enables an AV to drive autonomously from point A to point B on its own without a driver. This is where the first use of AVs will happen. More on this shortly.
Fully-autonomous driving enables a vehicle to drive everywhere on its own. This is still out in the future.
Another set of basics. There are several different types of technology platforms that enable autonomous driving.
The first type of platform is primarily camera-based. There are also other sensors, but cameras primarily are used to direct the vehicles based on what it sees, including lane markings, other vehicles, pedestrians etc. This is the primary platform used for semi-autonomous driving. It is good at highway and similar driving but challenged to negotiate turns and corners.
The second type of platform is very different. It uses lidar and predefined high-definition maps, in addition to cameras and other sensors. This technology platform uses lidar to position the vehicle precisely in the predefined map. It knows that it is three centimeters from the curb and the exact turning radius of the curb, so it can easily negotiate a turn. It is limited to driving in predefined areas (sometimes referred to as geo-fenced), but that is OK for the planned use. This technology is working today and has been safely tested for more than 10 million miles of driving.
The final set of basics to understand is that there are two very different primary markets for autonomous vehicles. There are also markets for trucking, shuttles, home-deliver, etc., but let’s focus on contrasting these two.
One market is privately-owned autonomous vehicles. This market is still five years away for many economic, technology, and use issues. All auto manufacturers have changed their strategies to make this a secondary market.
The next, and most important market, is Autonomous Ride Services (ARS). Essentially this is like Uber without a driver. This is an entirely new market opportunity, and it is where all AV companies are focusing their attention.
Autonomous vehicles will be introduced first as Autonomous Ride Services, using sufficiently-autonomous vehicles, based on lidar and predefined-maps.
In my recent book, I go into a lot more detail on this, but here is a summary. ARS companies will introduce large fleets of AVs in selected metropolitan areas. For the first few years, the metropolitan areas will be primarily in the South and West, and in states that permit and encourage AVs, such as Arizona, California, Florida, and Texas.
ARS companies will establish an operations center in each metropolitan area to oversee, dispatch, and maintain these fleets. As with ridesharing, they will use an app for riders to request and pay for a ride. The trips they provide will be limited at first to those locations and routes defined by its geo-fenced maps. But these will rapidly increase, covering all the most popular routes in that metropolitan area.
Riders will flock to use ARS not only for the novelty but for the cost. I expect the cost of an ARS trip will be approximately half of a comparable Uber trip; after all there are no driver costs.
Who will be providing these new ARS services? Waymo (Google) is by far the leader. It has driven more than 10 million miles without a driver and has simulated another 10 billion miles. It has tested its ARS without a driver (but with a safety driver on board) with real passengers for two years. It currently is testing with no safety driver in the car. It has a factory in Michigan to customize autonomous vehicles for this service, and it has contracts for the delivery of 100,000 car over the next few years.
Others expected to enter the U.S. market early include GM Cruise and Ford. Apple is the wild card here. It is secretly developing an AV for this market but hasn’t disclosed its plans.
So, when will ARS using autonomous vehicles be available? After updating all of my research, I still maintain my original forecast that the market will start in 2021 — a year from now. Depending on COVID-19 delays, Waymo will have limited ARS services later this year, but 2021 is when the market will start. From there, I expect it to grow rapidly from 30,000 vehicles in 2021 to 750,000 in 2025.
Michael E. McGrath is an experienced consultant, executive, author and board member. He wrote: Autonomous Vehicles: Opportunities, Strategies, and Disruptions.