We are just starting to enter the era of self-driving cars and we are going to be hearing a lot more about autonomous cars. One thing you might not be aware of is the classification of levels of automation – there are six of them. Here is an introduction to what each level of automation means; from the fully human-powered level of zero to the totally autonomous level of five.
This concept of levels of autonomy was originally published by the International Society of Automotive Engineers (SAE) in the year 2014. It was published as a part of its “Taxonomy and Definition for Terms Related to On-Road Motor Vehicle Automated Driving Systems” report, where the industries belonging to that body began to develop a language that could be used to discuss this growing field.
According to the report, there are six autonomy levels of autonomy that automakers will be achieving in the future of their car fleet. In September 2016, the National Highway Traffic Safety Administration (NHTSA) abandoned its formal classification system that was released in 2013 and accepted the SAE standard.
According to the SAE, the levels of automation are outlined below:
Level Zero: No Automation
This is the simplest level to grasp, the one in which everything is manual and we are all quite used to it. In level zero, the driver keeps full control over all vehicle primary functions, such as the brakes, acceleration, steering.
The system cannot take over the driving but, if it exists, can still assist the driver via audible alerts when crossing a line, or by using a rear-view camera, etc..
Level One: Driver Assistance
In this level, the system can control speed or direction. The driver keeps his hand on the other functions and still has the total control of the vehicle.
Adaptive Cruise Control is a good example of level one autonomous self-driving. Depending on the vehicle ahead of your car, the system may decide to slow down to maintain safe distances.
In Level one, the responsibility for control is permanently retained by the human, who delegates some of the tasks to the system, and who must be able to regain control if the situation so requires.
Level Two: Partial Automation
At this level, the car can take control of speed and direction or the pedals and wheel. An example is the park assist: the system is responsible for temporarily driving the car, while the driver only supervises the operations. However, the driver must be very careful, and to regain control in case of failure of the system because its responsibility is always fully committed to the human driver. Tesla, for example, has been at this level since 2014.
Level Three: Conditional Automation
In this level, the driver fully delegates driving in pre-defined situations to the system. With this level of automation, one can imagine reading his messages quickly, or his newspaper while in the morning traffic.
The system knows its limits, and its “driver” must still be able to take back the control when the system asks for it. When the vehicle notifies the driver of its inability to handle a situation, the driver is expected to retake control. Self-driving cars with this autonomy level can decide how to respond to dynamic incidents on the road, and when to change lanes, but uses the human driver as a fallback system. This level introduces a level of ambiguity over who is responsible for the car when it’s automated if there is some type of incident. The human should be aware and in control theoretically but the car is sharing in the decision making. Most auto-makers want to get to level four and skip this level.
Level Four: High Automation
Level four autonomy is characterized by the ability of the car to drive itself completely under the right circumstances. However, if it runs into a problem that is beyond its control, it can ask for assistance from a human, but if human help is not forthcoming, it can park itself without endangering its passengers. Level four autonomous car can move without its owner and can also be driven by a human.
The level four autonomy has basically been accomplished in the context of specific driving situations such as in the parking lot. In this case, the system takes over all the functions independently without even requiring the presence of the driver and also takes responsibility for the control; the car can park in a parking lot and also picks up the driver when it is time to leave.
This is the level where a truly self-driving car begins. Google/Waymo’s test cars have been operating at this level for some years now.
Level Five: Full Automation
Here, the car is capable of driving in all circumstances; whether you’re driving or not has no impact. In this level, the system takes control over all functions of the car.
This level of autonomy is distinguished from the previous level by “machine certainty,” a concept that allows the system not to execute a human order if it is considered abnormal, inconsiderate or dangerous or to take the initiative based on the measurements of its sensors.
In some cases, automotive systems can, therefore, engage in unsolicited moves by the driver or even refuse to carry out a movement that would endanger the vehicle or its passengers (for example, opening the door on a highway).
Just a few years ago self-driving cars seemed like futuristic fiction but with recent progress by the likes of Google, Uber and many of the major car manufacturers, the future may be arriving much sooner than expected.
Tesla has stated their ambitions to get a level five autonomous car in the next few years. It will be interesting if they can make it there that quickly or not. To some extent, the evolution of these technologies will depend on the speed at which we agree to adopt them as individuals, as business and with the support of government regulators.