Defining the key stages of vehicle autonomy (2/6)

4 mins read

The push to develop fully autonomous vehicles has been fixed in the crosshairs of most automobile manufacturers for a number of years now.

This is being driven by the need to meet future mobility demands and improve the safety of road users, not to mention keep pace with a consumer base engrossed by the latest technological advancements.

Populations are growing, cities are becoming larger and more crowded, and road networks are increasingly congested. Add to this the undeniable fact that human drivers remain the cause of the vast majority of road accidents, and the need to increase the use of intelligent technology to help get us to our destinations becomes indisputable.

The road to fully autonomous transportation, where we can simply hail a vehicle from a smartphone app, wait for it to arrive at our door, climb aboard, then watch the world go by until we arrive at our desired location, is a going to be complicated one. The sensor mechanisms, software and connectivity hardware required for a fully autonomous vehicle are highly sophisticated, expensive – and actually not even completely defined yet.

Though the engineering challenges are still being dealt with, the stages that will lead to fully autonomous vehicles have already been mapped out. In 2014, the Society of Automotive Engineers (SAE) produced an outline of the six levels of automation, with its “Taxonomy and Definitions for Terms Related to On-Road Motor Vehicle Automated Driving Systems J0316” standards. These standards were updated at the beginning of this year to reflect certain evolutions that had taken place.

We have already passed through the first two levels of autonomous functionality. Level 0 of the SAE standards is simple. Here the vehicle is always controlled by the driver, and there are no assistance systems of any note involved. The driver is responsible for steering, acceleration and braking, monitoring the environment around them, and any dynamic tasks (such as changing lanes on the motorway, or overtaking another vehicle).

Level 1 was reached when limited driver assistance systems started to be integrated, but the onus here has still remained on the driver. Adaptive cruise control and lane departure warnings may offer some degree of help, but it is very much up to the human behind the wheel to stay alert. They must be completely aware of the environment around them and react to any changes (like speed limits and road conditions) or hazards as they appear.

The lines begin to blur

Beyond Level 1, the definitions utilized by different manufacturers can fluctuate slightly. One of the most talked-about car companies, Tesla, often claims its Autopilot technology as a system that offers autonomous driving functionality. It can cite numerous cases of people sitting in the driver’s seat of vehicles in transit without having any contact with the steering wheel or pedals. If you speak to engineers developing systems at other companies however, they will say that in reality Tesla’s technology exists in between Levels 2 and 3, rather than further up the scale.

In true Level 2 systems, multiple functions (such as some of the ones already described) are able to work together to provide what is referred to as partial automation. These systems can assist with steering and acceleration/deceleration, making use of adaptive cruise control, lane centering and suchlike to achieve this. The driver must always be vigilant in case of sudden changes to their surroundings, though, and will still be responsible for dynamic driving tasks, so their hands need to remain on the steering wheel.

It is not until we reach Level 3, and the emergence of conditional automation, that vehicles begin to take on the task of monitoring the environment, and vehicle occupants can start to relax a little more. At this stage the onboard systems, such as traffic jam assist, can access contextual data and control acceleration, steering and braking accordingly. So for example in motorway traffic, the vehicle can maneuver without human intervention. The caveat is that the driver always has to be ready and able to intervene should the technology not be able to cope with the driving scenario presented.

All systems go

Things really get interesting when we reach Level 4, as pedals and a steering wheel become optional, and the necessity for driver intervention decreases dramatically. Here the vehicle will be in a position to accelerate, brake, steer, monitor its surroundings and (should the driver ignore requests to intervene) even perform dynamic driving tasks to maintain road user safety – coming to a complete stop if necessary.

The ultimate goal for autonomous vehicle technology is a car that is capable of maneuvering in all driving modes, no matter what events it faces. When this happens we will finally have arrived at Level 5. The onboard systems involved will be able to execute steering, acceleration and braking activities, monitor the surroundings and perform all ongoing dynamic driving tasks no matter what environmental or road conditions the vehicle encounters, with zero human intervention.

The final two levels of automated driving systems rely heavily on an extensive sensor suite being incorporated into the vehicle. This will have the capacity read and react to anything that occurs within the vicinity of the vehicle. As we shift from Level 4 to Level 5, the technological footprint will expand further to include high-speed wireless communication protocols. Next-generation networks supporting this will need to be rolled out, so that ultra-low latency data can be taken from other vehicles and road infrastructure (smart traffic signals, camera equipment, etc.), as well as augmenting information from high-definition maps.

The definitions of the numerous levels of autonomous vehicles give a marker of what to expect as each stage is attained. The meanings and descriptions for every one of the levels yet to come are almost certain to be adjusted as the supporting hardware/software develops. Likewise, the timeframes in which each should be expected is far from being totally fixed. The whole concept of fully autonomous driving should be viewed as having some fluidity, growing and adapting – just as will the technology on which it is based.