The idea behind vehicle autopilot is to have a car that drives itself, makes decisions on the fly, and essentially replaces the need to have a human driver behind the wheel. Can the vehicle detect traffic signals? Does it detect lane changes? Can it self-drive on the freeway? These are only some of the factors we analyze when assessing autopilot capabilities in electric vehicles.
The idea behind vehicle autopilot is one day the cars are supposed to be self-driving. The best vehicle autopilot systems today only still require the driver to be in control at all and only provide assistance in limited situations.
The society of automotive engineers has devised a framework for self-driving automation. As of today, the vehicle autopilot systems available in consumer cars are at best a level 2.
Looking at the self-driving framework three questions are asked:
The society of automotive engineers devised six levels to describe how automated a vehicle autopilot system would be.
Level 0 represents no automation at all. On the other end of the spectrum, level five represents that vision of a fully self-driving car that can handle all situations, thus replacing the human driver.
As of today, vehicle autopilot systems are at level 2, which still requires the human driver to control the vehicle and pay attention at all times.
Most current vehicle autopilot systems found in new cars are level 2. What level 2 represents is a vehicle that has semi-autonomous steering, accelerating, and braking. The vehicle autopilot systems typically work only on freeways and only in certain situations requiring the human driver to be attentive and ready to take back control of the vehicle at a moment’s notice. Therefore with a level 2 vehicle autopilot system, the human is responsible for driving the vehicle.
More economical vehicles or older electric cars may have level 1 driving automation as their vehicle autopilot system what a level 1 system would require the driver to be responsible for the vehicle at all times. For example, either steering or speed control may be semi-automated but not both at the same time. Adaptive cruise control systems found in cars are an example of level 1 automation as an example.
How vehicle autopilots visualize the world around them is through the use of sensors. These sensors are usually a combination of cameras and radar. There is another form of sensor called Lidar used on the Google Waymo vehicles, but this technology is too expensive for consumer passenger vehicles.
Car companies today use cameras and radar to make a rendition of the world and identify other vehicles, pedestrians, and other important markers. With cameras particularly during rain and snow, the technology works poorly, leading to vehicle autopilot systems being used primarily during clear weather.
Autopilots today used in passenger electric vehicles are typically a level 2 driving automation. This means the vehicle has the ability to steer, accelerate and brake in certain situations with required constant human supervision as the system can at any time pass back driving to the driver (think of this as a driving assistance feature). In order to determine and rank which autopilot systems are better than others, we use eight autopilot features.
Lane centering or auto-steering is the ability for an electric vehicle to be able to stay within a lane while on autopilot. Lane centering also has the ability to steer around curving roads. Current lane-centering systems rely on visible lane markings to work. If lane markings are obstructed or covered lane-centering may not work. We rate lane-centering on a five-star scale with five being best and one being worst. Also, weight lane-centering is higher than other autopilot features.
Another criteria we look at is if the carmaker includes autopilot or charges an additional fee. If the autopilot is included we consider that in our calculations over an equivalent system that costs an additional fee.
Adaptive Cruise Control is the ability of the autopilot system to accelerate, brake, and keep a distance away from other vehicles while driving. We look to see if adaptive cruise control is included as part of autopilot. Any level 2 autopilot system would require adaptive cruise control to function.
Being that current autopilot systems require the human driver to be alert and ready to take back control in the event the autopilot system passes back driving functions driving attention monitoring becomes an important feature.
Driver attention monitoring usually is looking that the driver has their hands on the steering wheel. Some driver monitoring systems look at where your eyes are looking to also determine driver attention. We look to see if the autopilot system has driver attention monitoring and if so we factor it in when evaluating an autopilot system.
Autopilot systems would typically be used in street driving throughout a city or a freeway. Freeway driving is the easier of the two scenarios, as the electric vehicle is typically driving on a straight road with minimal turns, no interceptions, and more predictable driving conditions. We look to see if an electric vehicle autopilot system can be deployed for freeway driving.
Autopilot traversing city streets is a more complex task and as of today one that autopilot systems are grappling with. Street driving requires autopilot to be able to understand streets, intersections, and more complex driving situations. With autopilot systems, if they are not able to operate on city streets and cannot handle every scenario, you will see a greater instance of the system passing back driving functions to the human driving.
We look to see if the autopilot system can operate on city streets and if so we factor it into our autopilot rankings.
One determinant that can assist the autopilot system is whether it can read traffic signals. If the vehicle can identify and understand the meaning of stoplights, caution signage and stop signs we factor that into autopilot rankings.
The ability for an autopilot system to make lane changes is a feature we look for. Autopilot lane change has the ability to see if there is a vehicle in the lane or if the lane is open. In the event, the lane is open the electric vehicle autopilot system would make a lane change on its own. We look for automated lane changes as criteria to determine when ranking autopilot systems.