Autonomous Driving Will production cars ever reach Autonomy Level 5?
While a few companies have beta-tested fully driverless cars on public roads, it remains an open question whether such vehicles will ever really be a common sight, at least in the near future.
The Society of Automotive Engineers (SAE) has defined graduated levels of autonomy for cars with self-driving capability to be rated at. The highest rating, Level 5, means that a car requires no driver at all and can convey a passenger from a starting point to a destination without any human intervention.
So far, companies like Waymo, Tesla, Apple, and others have tested cars rated at SAE Level 5 on public roads in multiple states in the U.S. where such activity is permitted. But these firms have not yet made such cars available either for sale or as components of "Robo-Taxi" fleets.
In a few cases, Level 5 functionality is being delivered on a limited basis in closed areas such as airports and parking lots. But providing true Level 5 functionality in open-range spaces and cities may be much harder to realize due to the unpredictability of local roads, weather, and traffic, among other factors.
Why Level 5 is the "Holy Grail" of autonomy
Delivering Level 5 functionality has been both a dream and a fleeting promise from car manufacturers ever since the 1930s. Automakers have tempted the consumer with promotional films, brochures, and ads teasing completely automated self-driving functionality as a feature sure to appear in a future that was just 20 years away. Thus, consumers of the 1930s were persuaded that by the 1950s, such cars would be seen on local streets. In the 1940s, the timeline was set for the 1960s, and in the 1950s, it was the 1970s, and so on.
But at some point—likely in the early 2000s—the timeline shifted from being 20 years away to 10 years, and finally, just five years. And this time, the predictions came true. Companies like Waymo (Google), Apple, and the newly born Tesla aggressively began developing self-driving cars that slowly pushed their way from Autonomy Level 0 (no automation) to Autonomy Level 4 (driver present, but mostly as a passenger).
Waymo, in particular, racked up more than 32.1 million kilometers (20 million miles) in testing, while the others also made significant progress.
The goal was always to make a completely driverless car (eventually lacking a steering wheel and any manual controls). This car could be summoned remotely and would pick you up at a location and drive you to your destination without assistance.
Such vehicles could be shared and could save countless kilometers and hours of commuter distance and time through increased journey efficiency. By extension, this would also result in lower traffic density.
What stands in the way of Level 5 autonomous cars?
Despite great strides being made, however, universal Level 5 driving remains more of a "wish-list" goal than a reality. The reason is simple: real-world driving scenarios can be much more complex than current navigation technology is able to process.
Consider two scenarios: the first is a highway with safety dividers between sides, clearly demarcated lanes, reflective road markers, speed limit signs, and high visibility.
For self-driving cars equipped with a variety of camera, radar, and LiDAR sensors, this first scenario is relatively easily handled. Proper distances can be maintained between cars, and the vehicle in question can be automatically kept centered in its lane.
Speed, braking, and acceleration are all relatively easy to control because every car on the highway is traveling at close to a uniform speed. Steering is easy because there are no sharp turns or sudden obstacles to avoid.
Now consider a second scenario: a two-lane country road that has no painted dividing line, with multiple asymmetrical intersections, and random traffic. This road also has poor visibility due to frequent bad weather, and unpredictable moving obstacles, such as animals or pedestrians that cross the road in unexpected places.