“Look, ma! No hands!” Science fiction has shown us countless examples of self-driving cars. How close are we to this phenomenon? (Image courtesy of NHTSA.)

Growing up, I was a huge fan of the Knight Rider TV show. While the prospect of a Pontiac Trans Am going really fast and performing stunts will never stop appealing to young people around the world, what made it even cooler was the fact that the car, K.I.T.T., could talk and think. It was a fully autonomous car that—aside from its penchant for deadpan sarcasm—was capable of making its own decisions, holding conversations and driving itself.

As I get older, I still watch clips of the show for nostalgia and find that the premise of an autonomous car continues to hold well (the same cannot be said of David Hasselhoff’s mullet though). Self-driving cars are a fascinating concept for many because they are convenient, they are current as we progress more and more towards automation, and they are cool. So very cool. Whether it’s because of Knight Rider, Minority Report or Transformers, the thought of your car just driving itself to your destination while you sit back and relax is appealing.

So, if self-driving cars are convenient and current and cool—why aren’t they here yet?

The Six Levels of Self-Automation

The Society of Automotive Engineers (SAE) has divided driving automation into six different categories—Level 0 to Level 5—that have been adopted by the U.S. Department of Transportation.

(Image courtesy of Synopsys.)

Level 0, Level 1 and Level 2 Automation

The vast majority of vehicles on the road today fall under Level 0. At this level, cars are entirely manual, requiring the human operator to control all aspects of driving—from acceleration and deceleration to steering.

Level 1 is the first step towards automation and entails technology that assists the human operator in driving the car. This technology includes cruise control that helps maintain a safe distance between cars, and parking assist technology that alerts the driver to possible collisions and allows them a better understanding of spatial positioning through sensors, side and rear cameras, etc. These technologies fall under Level 1 because the driver is still ultimately responsible for steering and controlling the car’s speed.

At Level 2, the vehicle is capable of achieving partial driving automation. At this level, the vehicle can steer as well as monitor its speed. Vehicles like Tesla and Audi A8 are examples of Level 2 automation. That said, the driver is still required to take control of the car at a moment’s notice. Tesla’s Autopilot is an example of Level 2 automation, funneling information from its eight cameras and 12 ultrasonic detectors into its onboard computer that is constantly analyzing this data in a way that keeps the driver informed of their surroundings. Through Autopilot, the car maintains its speed based on the speed of surrounding cars, responds to traffic lights, automatically steers to stay in a selected lane, and even parks in parallel or perpendicular. In fact, it can be summoned from its parking spot to navigate to you or your location of choice, all the while maneuvering around objects to get to you—yes, just like the Batmobile!

Tesla’s autopilot system keeps the driver constantly aware of their surroundings through sensors and a 360-degree field of vision. (Image courtesy of Tesla.)

General Motors’ Super Cruise and Ford’s BlueCruise operate on a similar principle by using sensors and cameras to provide a 360-degree world view to the driver. BlueCruise differs slightly in that it allows the driver to take their hand off the wheel as long as there are clearly defined road signals, road marks, and the driver is still paying attention. (Onboard cameras keep track of the driver’s eyes and face for signs of “dozing off”, prompting an alert.) Yet, for all this technological advancement that seems straight out of sci-fi, Level 2 still only counts as partial automation because the driver needs to stay alert and take control of the car.

Level 3 and Level 4 Automation

While the transition from Level 2 to Level 3 may appear gradual, it is a significant leap. So much so, that as of today, only one car qualifies as Level 3: the 2019 Audi A8L. (Due to the fluctuating regulatory processes in the U.S., the A8L is still considered a Level 2 in the U.S.)

What marks the A8L as a Level 3 car is its ability to make decisions based on the information it receives. For example, it boasts the Traffic Jam Pilot which—after being selected by the driver—assumes control of the vehicle in traffic jams, taking over responsibility for monitoring safe distance, speed and steering as long as the speed of the surrounding cars is below 60 km/h (ergo, traffic jam).

Naturally, there are areas in a city where the driver is required to drive below 60 km/h. Does that mean that any time the car is this slow, the traffic jam pilot will be activated? No. Through its 24 sensors, the A8L detects when it’s on a highway or a multi-lane road with clear markings. (It can recognize when it’s on a highway by detecting the separators placed between highway roads.) Ultimately though, like with Level 2, Level 3 automation still requires the driver to remain alert and make decisions on the fly. This process by which a car with self-driving capabilities can only do so in certain speed limits and traffic conditions is known as “geofencing.”

The control unit of the Audi A8L. With a total of 24 sensors and a 360-degree view, the A8L provides a rich tapestry of information to not only inform the driver but also make driving and safety-related decisions. (Image courtesy of TopSpeed.)

At Level 4, the key difference is that unlike Level 2 and 3, the automation is capable of making decisions in the event of system failure or other issues. That is, the car is self-driving and the human driver does not need to intervene in most driving situations. However, there are still limitations in place that allow the human to override and take control of the car should the need arise. Moreover, even at Level 4, geofencing is still in effect, though the exact value will differ from company to company. Presently, Level 4 is not being used for commercially sold cars. Instead, efforts are in place to use Level 4 autonomous cars as taxis and shuttle services in certain sections of a city.

Phoenix-based Waymo is one of the first companies in the world to offer taxi services with Level 4 driverless cars. Dubbed Waymo One, these self-driving cars operate much like Uber or Lyft, where the customer hails them via an app. The car (or SUV) will pick and drop customers at select locations inside the state of Arizona, and more recently, in San Francisco. Another facet of geofencing is that the Waymo One can only take roads with a maximum speed of 73 km/h. The customer can add up to five stops in a single trip, and other than needing to wear a seatbelt at all times, does not have to do anything else. By January 2020, Waymo One had completed over 20 million miles of autonomous driving.

The Waymo One fleet includes Toyota Priuses, Lexus SUVs, and fully autonomous Chrysler Pacifica minivans. (Image courtesy of Waymo.)

Yet, Waymo One’s endeavors are not without issues. In May 2021, Waymo One had a famous (and hilarious) malfunctioning episode that was livestreamed by an unfortunate client. The driverless car was rendered confused because of driving cones and kept circling a particular road, eventually dangerously blocking the road. When a team of Waymo troubleshooters arrived on the scene for assistance, the car promptly scuttled off as if afraid of getting in trouble. (“I've never seen a robot have an anxiety attack before lol,” said one YouTube commenter.) Waymo has since claimed to have resolved the issue and insists that this was a one-off incident.

France-based NAVYA offers a similar traveling solution, this time with shuttle buses. With a capacity of 15 people, the Autonom Shuttle Evo is a fully electric shuttle that can drive at 40 km/h and has been deployed in over 20 countries.

Over 180 units of the Autonom Shuttle Evo have been sold around the world. (Image courtesy of NAVYA.)

In Canada, Magna hopes to bring Level 4 autonomy to vehicles through its MAX4 technology. MAX4 uses IoT that combines light detection and radars (LiDAR), cameras and ultrasonic sensors. The IoT is then incorporated into hybrid and electric cars without impacting their appearance.

MAX4 enables Level 4 capabilities in vehicles traversing urban roads and highways. The entire system is button-operated, meaning that when the driver wishes for the MAX4 system to take over driving, all they have to do is set the destination and press the engage button. The MAX4 system can be deployed on different types of vehicles, from standard cars to SUVs.

“Many of the autonomy challenges the industry is facing can’t be handled in isolation,” said Magna’s CEO, Swamy Kotagiri. “I am confident that the many strengths brought forward by each partner will contribute to the delivery of a self-driving platform the market can leverage.”

Similar advances have been made in China, where 3000 “robot taxis” are set to be built by tech giant Baidu in the next decade. Over 12 million kilometers of autonomous tests have already been conducted by Baidu as it plans to launch the taxis in over 30 cities across China.

Level 5 Automation

At Level 5, the human driver is not required to drive the car at all. Geofencing will also be eschewed altogether, leading to self-driving cars that can traverse all sections of the city and all terrains. Many expect Level 5 vehicles to not even have steering wheels, accelerators and braking pedals as all aspects of “driving” are automated. The bad news, however, is that we are nowhere close to this level of automation, which means that Knight Rider fans still have a ways to go before they can whisper “K.I.T.T, come on!” into their watches and have a Pontiac Trans Am burst into the scene.

But we can dream, right?