Autonomous Vehicles: Where Are We Today?
Although a number of useful driving aids have made their way onto the market, true self-driving cars are still in the future.
Cars that drive themselves. The promises made by the developers of autonomous vehicles are simple and beguiling. Yet, after almost 10 years of hype and headlines, it turns out that the challenges have been much more difficult than anticipated.
A Cruise autonomous car. Image used courtesy of Cruise
On the Level
The Society of Automotive Engineers (SAE) has described six categories assessing vehicle autonomy, from Level 0 (no automation) to Level 5 (full automation).
Level 0: No automation. The human driver can control the vehicle at all times.
Level 1: Driver assistance. The vehicle can control some aspects of driving, such as steering or braking, but the human driver must still be alert and ready to take over.
Level 2: Partial automation. The vehicle can control most aspects of driving, but the human driver must still be ready to take over in certain situations, such as when the vehicle encounters an unexpected obstacle.
Level 3: Conditional automation. The vehicle can drive itself in most situations, but the human driver must be available to take over at all times.
Level 4: High automation. The vehicle can drive itself in most situations without human input, but it may require human intervention in rare circumstances.
Level 5: Full automation. The vehicle can drive itself in all situations without any human input.
Level 1: Here Today
Level 1 autonomy is the most basic level of automation. Many vehicles today have adaptive cruise control, which automatically adjusts the vehicle's speed to maintain a safe distance from the car in front, and lane keeping assist, which helps the vehicle stay in its lane.
An example of Level 1 autonomy includes the 2023 Toyota Camry with Toyota Safety Sense 2.5+, which provides Dynamic Radar Cruise Control and Lane Departure Alert with Steering Assist. The Toyota also features Pre-Collision System with Pedestrian Detection, which uses cameras and radar sensors to determine whether a crash with another vehicle is likely or pedestrians are in danger. It then uses the information to alert the driver. If the driver doesn’t respond quickly enough, the system applies the brakes automatically to avoid or reduce the severity of a collision.
Another example of Level 1 autonomy is the 2023 Ford Fusion with Ford Co-Pilot360. In addition to lane keeping and Pre-Collision Assist with Automatic Emergency Braking, the Ford has options for Adaptive Cruise Control with Stop & Go and Lane Centering, Post-Collision Braking, and Evasive Steering Assist.
Level 2: Partial Automation
Level 2 autonomy augments Level 1 features by adding technologies that allow the vehicle to steer, accelerate, and brake itself in certain situations.
One example of Level 2 autonomous technology is employed by the Tesla Autopilot. Available on all Tesla models, it features traffic-aware cruise control, lane centering, automatic lane changing, semi-autonomous navigation on some freeways, and self-parking. It also allows the driver to summon the car from a parking spot or a garage. Eight cameras provide 360 degrees of visibility at up to 250 meters of range while vision processing is accomplished using a Tesla-developed computer using a neural net.
Cadillac is another company that has embraced Level 2 autonomy with its Cadillac Super Cruise system.
Cadillac’s Super Cruise allows hands-free driving. Image used courtesy of Cadillac
Super Cruise uses real-time cameras, sensors, GPS data, and laser radar (LiDAR), along with precision map data to determine the vehicle’s position and to help it stay centered in the lane. It allows hands-free driving along with adaptive cruise control and lane-changing on demand and can even be used when towing a trailer.
BMW Highway Assistant has a host of Level 2 features, including road sign recognition, cruise control, distance control, and assists for emergency braking, steering, lane guidance, hill start, speed limit, lane-changing, and parking.
Level 3: Conditional Automation
Level 3 autonomy allows the vehicle to drive itself in most situations, but the human driver must be available to take over at all times.
One example of a Level 3 autonomous vehicle is the Mercedes-Benz Drive Pilot. Mercedes has offered Level 2 autonomy for several years, including the “Distronic” Active Distance Assist that can automatically maintain a preselected distance to vehicles ahead on all road types, and active steering assist that provides lane-following support in a speed range of up to 130 mph. Active lane-keeping assist employs a camera to detect lane markings and lane edges at speeds that range from 37 to 155 mph to help the driver stay in their lane.
Examples of driver assistance technology. Image used courtesy of Mercedes-Benz
In December 2021, Mercedes introduced its Level 3 Drive Pilot, which met the requirements of the international standard UN-R157 for Level 3 conditionally automated driving. Mercedes-Benz is initially offering its Level 3 system only in Germany, where 13,191 kilometers of highway have been approved for automated driving. Drive Pilot allows the system to take over driving in heavy traffic or highly congested areas at speeds up to 36 mph (60 kph) and allows the driver to disengage from the vehicle control requirements.
The Level 3 system has been available in Germany for S-Class and E-Class models since May 2022. Mercedes-Benz has also become the first automaker to receive equivalent approval in the U.S. in California and Nevada, and the technology will be available on the U.S. market as an option on 2024 S Class and E Class models.
Drive Pilot uses the existing surround sensors of the driving assistance package, adding additional sensors that Mercedes-Benz considers indispensable for Level 3 driving. These include LiDAR and a camera and microphone in the rear window that can be used to detect flashing lights and sirens from emergency vehicles. A highly accurate positioning system, LiDAR uses data obtained from satellite navigation and matches it with sensor data and data from an HD map to help guide the system. In addition, data collected by LiDAR, camera, radar, and ultrasonic sensors provides real-time information on road conditions, traffic, and road signs.
Level 4 and Level 5: High and Full Automation
The development work on Level 3 autonomy has made clear that autonomous vehicles need to be able to perceive their surroundings accurately and make decisions in real time. This will become even more critical as autonomy moves into Level 4 and ultimately to Level 5. Vehicles will need to be able to handle unexpected situations, such as bad weather or road construction. System performances can be degraded by weather conditions (heavy rain, snow, smoke, fog), bright lights (direct sunlight, oncoming headlights), road conditions (steep, narrow, winding), or temperature extremes.
Level 5 autonomy will have its specific challenges:
The need for a robust mapping system. Level 5 vehicles will need to be able to map their surroundings in great detail, including the location of every object, such as cars, pedestrians, and buildings. This is a challenging task, and it is unclear how it will be accomplished.
The need for a reliable communication system. Cars will have to communicate with other vehicles on the road and with infrastructure, such as traffic lights and road signs. This is necessary to ensure that they can coordinate their movements and avoid accidents.
The need for a high level of trust. Autonomous vehicles classified as Level 5 will need to be trusted by the public before they can be widely adopted. This will entail extensive testing and public education.
It isn’t just automakers who are working on Level 4 and Level 5 autonomy. Several independent companies are developing robotaxis—vehicles that operate like regular taxicabs but without drivers. Level 4 and Level 5 autonomous vehicles for robotaxis provide several benefits:
They can operate 24/7, which can help to improve traffic flow and reduce congestion.
They can be used to transport people who are unable to drive themselves, such as people who are elderly or have disabilities.
They can be used to provide shared transportation services, which can help to reduce the number of cars on the road.
They can be used to collect data about traffic patterns and road conditions, which can be analyzed to improve the efficiency of the transportation system.
Among the companies developing a robotaxi system is Waymo, a subsidiary of Alphabet Inc. Waymo's robotaxis are operating in a limited capacity in Phoenix, Arizona. The vehicles are equipped with multiple sensors, including cameras, radar, and LiDAR. Waymo is also developing a robust mapping system and a reliable communication system for its robotaxis.
A Waymo robotaxi. Image used courtesy of Waymo
Another company developing the robotaxi concept is Cruise. It is a subsidiary of General Motors that is developing autonomous vehicles. Cruise's robotaxis are currently operating in a limited capacity in San Francisco, California. The vehicles are equipped with a set of sensors similar to Waymo's robotaxis.
These are just a few of the companies that are developing Level 4 and Level 5 autonomous vehicles for robotaxis. The development of this technology remains in early stages, but it has the potential to revolutionize transportation.
Autonomous Vehicle Challenges Ahead
It turns out that moving from Level 3 autonomy to Level 4 and Level 5 is harder than expected. Some of the problems that have been encountered include:
Sensor errors. Sensors can be fooled by bad weather, glare, or other road conditions. This can lead to the autonomous vehicle making mistakes, such as crashing into something or running a red light.
Software bugs. The software that controls the vehicle is complex and can contain bugs. These bugs can also lead to mistakes.
Hacking. Autonomous vehicles may be vulnerable to hacking. If the vehicle is hacked, the hacker could take control of the vehicle and cause it to do something dangerous, such as crash into another vehicle or cause a traffic jam.
Public acceptance. Some people are not comfortable with the idea of riding in an autonomous vehicle. They worry that the vehicle will make mistakes or that they will not be able to trust it.
Despite these challenges, the development of autonomous vehicles is moving forward but is taking longer than had been expected. Not the least of the problems is one of cost. Level 4 and Level 5 autonomous vehicles will likely be more expensive than traditional vehicles. This is because they are equipped with more sensors and software, and they need to be more robust to handle unexpected situations. Above all, they must be reliable and safe, and reaching that stage will take a lot more time and money.