Automotive AI places a strong emphasis on developing autonomous vehicles. However, something many people tend to forget is that autonomy isn’t an ‘all-or-nothing’ concept. There are quite a few steps between a vehicle with limited automation tools and a self-driving car. Autonomous driving systems can be thought of as devices in a network that operate free from human interference.
For example, last year, Caterpillar autonomously hauled 1.2 billion tons in its off-highway mining trucks. This is one of many illustrations that large assets that move heavy payloads can work efficiently and safely without a human operator.
Not all autonomous driving systems are created equally, and understanding the different levels of autonomy is crucial when working in automotive technology.
With over 85 years of experience in electronics, Morey is a thought leader in connected products. Our mission is to share our vast knowledge with others so we can empower our partners to fulfill their goals using smart, integrated solutions. This article sheds light on levels of driving automation, as well as the state of autonomous driving systems.
What entity defines the levels of driving automation?
With a team of over 100,000 technical experts, SAE International is a leader in engineering. Its mission is to promote safe, clean, and accessible solutions in the aerospace, automotive, and commercial vehicle industries. Back in 2014, SAE International released its taxonomy of driving automation systems, which has become the standard for assessing levels of autonomous vehicles.
The levels of automation are determined based on which Advanced Driving Assistance Systems (ADAS) are available in the vehicle. ADAS uses cameras, radar, LiDAR, vehicle-to-X communication, GPS, and mapping data to provide more extensive assistance to the driver.
Examples of ADAS
- Lane Departure Warning (LDW): LDW warns drivers about vehicles crossing into their lane using cameras mounted inside the vehicle, radar sensors on the sides of the road, GPS data, and information from nearby cars.
- Blind Spot Detection (BSD): BSD detects objects moving around the side of the car and alerts the driver via visual indicators and/or audio cues.
- Adaptive Cruise Control (ACC): ACC adjusts speed based on how close another vehicle is behind you, which helps keep traffic flowing smoothly by reducing tailgating.
- Forward Collision Alert (FCA): FCA sounds an alarm when the front bumper of the car gets too close to the rear of the car in front of you.
- Automatic Emergency Braking (AEB): AEB activates the brakes if the car senses that you’re about to hit something.
- Pedestrian Recognition System (PRS): PRS uses cameras and radar sensors to identify pedestrians walking along the street and alert the driver.
- According to SAE’s J3016 system, there are six levels of autonomy (starting at level zero and progressing to level five), which we’ll discuss below.
SAE J3016 Level 0 “Limited Driver assist systems and no automation technology.”
This level refers to a vehicle that is not equipped with any driving automation technology. The driver is entirely responsible for operating the vehicle’s movements, which include actions like steering, braking, accelerating, parking, loading, and unloading.
At this level, driving automation technology is not installed; however, driver support technologies can be installed in the vehicle. These driver support systems include automatic emergency braking, brake assist, stability control, traction control, and blind spot warning. These technologies assist drivers by providing momentary support in specific situations, but they cannot operate the vehicle. The first variations of these systems appeared in the early 90s, but their production did not appear until the late 1990s.
In 1997 Toyota introduced the first production Laser Adaptive Cruise Control on its second generation Celsior. By the early 2000s, the Celsior model would feature most of the basic Driver Assist Systems which now come standard on most vehicles today.
SAE J3016 Level 1 “Driving Assistance”
The defining characteristic of this level is that the operator has at least one driving support system that assists in core actions such as braking, accelerating, or steering. The operator is completely responsible for the vehicle’s movement but has advanced levels of driver support technologies. A common example is adaptive cruise control (ACC), which maintains a safe following distance between the vehicle and the traffic ahead without the driver intervening. Another key technology at this level is steering assistance features lane centering technology, a rudimentary version of a Lane Keep Assist System (LKAS) .
Note: A vehicle with both LKAS and ACC will fall under the category of Level 2.
SAE J3016 Level 2 “ Multiple ADAS Features”
The staple for classification as a Level 2 system is multiple Advanced Driver Assistance Systems(ADAS) features. ADAS systems are capable of taking over the 3 major components of operation; steering, braking, and acceleration. The defining characteristic of this level is its limitation of use and the requirement of a fully alert operator who may take over full control of the vehicle at any time. Highway Driving Assist (HDA), installed in Genesis, Hyundai, and Kia vehicles fall in the category of Level 2 driving automation technology as it requires the driver to hold the steering wheel while HDA actively steers, accelerates, and brakes when traveling on highways. Slightly advanced versions of level 2 ADAS technology have been available for 7 years since the release of Tesla’s Autopilot in 2014. More recent personal vehicle OEMs (Original Equipment Manufacturer) examples include Super Cruise from Cadillac and BlueCruise from Ford. Current versions of these technologies allow the operator to remove their hands from the steering wheel on approved highways.
SAE J3016 Level 3 “Conditional Autonomous Driving”
The quantum leap from the Level 2 ADAS System to the Level 3 Conditional Autonomous System is so vast that no Level 3 systems are legal to use on American roads. A Level 3 autonomous driving system employs various driver support technologies and applies Artificial Intelligence (AI) in an Artificial Neural Network (ANN) to make decisions regarding vehicle movement in a changing environment. In Level 3 vehicle autonomy, an operator is still needed and must be present to take control in the case of system failure or an emergency situation. A major difference between Level 2 and Level 3 is the absence of a constant need for supervision of the ADS technology. In Level 3 and continuing on to Level 5 an operator is not needed when an ADS feature is engaged.
On March 4th, 2021, Honda became the first automaker in the world to sell an approved Level 3 ADS. Only 100 Honda Legend Hybrid EX’s were available for lease with the Sensing Elite package featuring a Traffic Jam Pilot Assist (TJPA) system. As of this writing, Japan’s Honda Legend is still the only country and car with a Level 3 ADS legally operating on public roads, despite Germany in 2017 becoming the first country to create the legal basis for the use of these Level 3 systems. The UN has since followed with the approval of UN R157. In Q4 of 2021, German automaker Mercedes-Benz announced its Drive Pilot as the first Level 3 system in the world to achieve regulatory approval on an international level.
SAE J3016 Level 4 “High Driving Automation & Full ADS Capabilities”
Level 4 ADS systems will require no human interaction and is the first SAE level where vehicles can be manufactured absent of operator controls. All movement and operations of a level 4 autonomous vehicle will be handled by the vehicle’s AI and its Artificial Neural Network (ANN). The major difference between a level 3 and a level 4 autonomous vehicle is the vehicle is programmed to intervene in the event of a system failure. However, not all scenarios and conditions are known and prepared for handling. A Level 4 autonomous vehicle is only able to operate if all the required conditions are met. Common conditions would include specified geographical area; vehicle being calibrated within a calibration interval or the overall health of the system, and all components being within predetermined thresholds.
The Autonomous Mining Trucks and Aerospace Auto-Pilot Are Level 4 AVs. Learn About Our Partnership With Caterpillar, Our Rugged Fully Customizable Gateways, or our history in Aerospace.
There are many conversations taking place regarding Level 4 autonomy. One commonly held belief amongst experts is the transportation industry skipping Level 3 autonomy altogether. Some experts believe OEM’s (Original Equipment Manufacturer’s) and governing bodies should focus their resources on developing the ecosystems necessary to support Level 4 autonomous vehicles. With that being said, most major OEM’s (Original Equipment Manufacturers) only have concept cars capable of Level 4 & 5 autonomy, and very few companies have prototypes currently in testing. The rubber met the road for level 4 autonomy in October of 2017 when Waymo started its Early Rider Program. Waymo launched a fleet of Level 4 autonomous vehicles operating as taxis but were limited to a geofenced area of 100 miles around Chandler, Arizona. A few companies who are currently major players in this space are: (Hyundai & Boston Dynamics Pilot Program “Roboride”) (Daimler & Waymo “Freightliner Cascadia”)
SAE J3016 Level 5 Autonomy– Full Driving Automation
A Level 5 autonomous system is completely free from human intervention and is not confined by geographical location. Currently, there are vehicles and systems capable of operation at this level. Many major OEMs have projects publicly claiming their vehicles are capable of full Level 5 autonomy. Tesla Autopilot states “all new Tesla cars have the hardware needed in the future for full self-driving in almost all circumstances.” This raises the question, “If the hardware is here then what is missing?” All of the systems needed for full autonomy are present in a Level 3 autonomous vehicle. The main differentiator from Level 3 autonomy to Level 4 autonomy and on to Level 5 autonomy is in the programs set to run, keeping in mind their limitations.
Learn more about Machine Learning & Artificial Intelligence in Our Article On IoT!
The software component (Artificial Intelligence) of these autonomous vehicle projects is quickly gaining on the standard driving scenarios and the maneuvers necessary for effective operations within those particular conditions. Scaling is still needed with these systems from being able to operate on a well-marked highway to an AI capable of recognizing and navigating a scenario where an officer is redirecting traffic so that vehicles are going the wrong way down a one-way street lined with construction due to an accident.
Challenges of developing a level 5 Autonomous Driving System
The implementation of a Level 5 autonomous system will have to overcome a few complex problems before full implementation.
- Safety: The first hurdle is achieving a level of safety that matches and surpasses that of the average human. The data suggests this hurdle has yet to be passed. Level 5 autonomous vehicle technology will demand major infrastructure improvements and maintenance.
- Legal: Major legal battles over who owns the data from ADS will continue to determine who will be held accountable in the event of a system failure. Governing bodies and the public will have to come to a consensus on the morality and repercussions of a life and death decision made by an algorithm.
The Future of the 6 Levels of Autonomy
As it stands, most vehicles on the road today are level 0-2 in terms of the SAE levels of autonomy. However, there is much promise for the future of ADS, considering the leaps made by companies such as Tesla and Mercedes-Benz, among others progressing to level three autonomy and above. While safety and legal concerns remain, it’s clear that companies are working hard to find solutions such as more formalized regulations and additional safety testing.
So, what is the next step toward employing an ADS? If you’re considering taking your ADS to the next level of autonomy, consider partnering with Morey. As a leader in IoT and connected product engineering, our team brings a level of expertise that cannot be understated. Contact us today to begin our collaboration!
Learn more about the hardware and software that make up the technology in a Fully Autonomous Vehicle.