What is an autonomous vehicle?

Autonomous vehicles, also sometimes referred to as self-driving cars or automated vehicles, are a category of automobile where some or all of the driving choices are made without a human. An autonomous vehicle is "autonomous" because it is capable of using sensors to get a picture of its operating environment and make driving decisions based on that data without the intervention of a human operator.

Though some assume that autonomous vehicles simply self-drive themselves, The Society of Automotive Engineers (SAE) defines six classifications of autonomous driving ranging from low-level help for a human driver to complete autonomous control of the vehicle.

The levels of driving automation are:

Human assistance:

• 0: No Automation: Humans perform all driving functions
• 1: Driver Assistance: There is a single automated system assisting driving (for example, cruise control)
• 2: Partial Automation: The vehicle performs acceleration and steering, but the human still monitors all tasks and has the ability to take control of steering and acceleration at any time

Automated driving:

• 3: Conditional Automation: The vehicle performs almost all driving tasks on its own, but a human override is still in place and needed.
• 4: High Automation: The vehicle performs all driving tasks under most circumstances, and the vehicle is spatially aware using a technique called geofencing. Human override of the vehicle is still possible, but not needed.
• 5: Full automation: The vehicle performs all driving under all circumstances and conditions. No human interaction or attention is required, and the vehicle may not even have a human in it at all.

How autonomous vehicles work can vary significantly between the various different technologies used to create them. At a basic level, the vehicle creates a map of its surroundings using an array of sensors giving it information about the roadway around itself and other vehicles in its path. This information is then interpreted by a complex machine-learning algorithm which then derives a set of actions for the vehicle to take. These actions are constantly changed and updated as the algorithm receives new information about the vehicle’s surroundings.

Although automated and autonomous are often used interchangeably, when talking on a technical level, they have slightly different meanings.

• Autonomous: The vehicle can not only drive by itself, but it can make choices that are truly its own. For example, when asked to take you to your destination, and you have a health emergency, it could take you to get care without being asked.
• Automated: An automated vehicle is able to automate a task, in this case driving, but it will always attempt to complete the task put in front of it. For example, an automated vehicle could be instructed to drive on the freeway, but could not course correct automatically if the driver got lost.

For this reason, the SAE lists the levels of self-driving as "automated" not "autonomous."

There are significant challenges for automated and autonomous vehicles, from technological, political, perceptual, legal, and even philosophical.

• Technological: Although self-driving technology has been in cars arguably as long as we’ve had cruise control, there are still large technological barriers for getting level 4 and level 5 cars into mainstream use. These include
  • Multiple autonomous vehicles getting "confused" by each other’s radar and lidar signals.
  • Lowered driving accuracy during intense weather conditions such as heavy rain, ice, or snow.
  • Correct identification of all objects in and around the vehicle at all times.
• Perceptual: Although automated vehicles can be made safe, the public perception of that safety is different than that of human drivers. While human drivers create a significant amount of injury and death in the act of driving, a similar amount of injury and death would likely be deemed unpalatable by the public in self-driving cars. Thus, automated vehicles need to not match the safety of human driving, but surpass it.
• Legal: If an automated vehicle causes injury, death, or destruction of property, who is liable for that? Is it the owner of the vehicle? The passengers? The manufacturer? These are questions that do not yet have answers, either legislated or by precedent in the courts, and would be a major open question for any self-driving enterprise at scale.
• Political: The legal framework for autonomous vehicles has, for the most part, not yet been set, and varies wildly between jurisdictions. Certain proposed laws, such as a distance tax on autonomous vehicles could implement massive liabilities on any autonomous vehicle enterprise.
• Philosophical: Underlying all of these issues are the human decisions behind the autonomous vehicle’s algorithmic programming. Who the vehicle would choose to prioritize in an accident is a key consideration, for example.

A software-defined vehicle is a digitally connected automotive which enables advancements such as electrification, autonomous driving, and seamless connectivity. When vehicles are designed to be software-defined rather than software-enhanced, they have the opportunity to provide continuous updates long after the purchase date of the car.

As software increasingly defines and drives the technology we use every day – from your phone to your smart home - consumers expect their car to follow suit. When the technology in their vehicles doesn’t match the advanced technology they enjoy in other areas of their lives, consumers notice the gap. 

The current development process and timeline for cars is much slower than for consumer electronics–this is where the software-defined vehicle comes in. Automakers are looking to use software as the main driver of development over hardware. The automotive and software industries are coming together to innovate faster and enhance the vehicle’s full digital lifecycle.

With the move to automated driving, more and more vehicle offerings are becoming software-defined vehicles.

At a basic level, relieving human beings of the stressful and attention consuming task of guiding a  high-skill transportation tool to get where they need to go would be a huge benefit. Not only would people be able to regain the time lost to commuting, they would also be able to do so in an environment that was safer for both drivers and pedestrians.

But that’s just assuming taking the current model of personal ownership of cars and transposing self-driving onto it. If we rethink the entire way that transportation works, it could have massive benefits. For example, if people no longer personally owned cars, but had a stake in a pool of vehicles (either through a SaaS-like service, or an analog to public transit), then each person would only need to order a vehicle when they needed it and it would move onto another person as soon as the trip was done. This would increase the sustainability of transportation incredibly, and significantly reduce the amount of vehicles on the road. Additionally, large amounts of real-estate could be saved in parking, because each vehicle would always be efficiently in use.

In-vehicle technologies

Customer preferences are changing rapidly. In order to keep up, automakers are required to evolve with emerging technologies just as quickly. 

Red Hat® can help vehicle manufacturers get to market faster with solutions and supporting technologies that accelerate transformation safely with an open source, enterprise-grade, Linux®-based in-vehicle platform. 

Red Hat’s solutions enable the connected vehicle through digital prototyping, virtual testing, and developing services in support of advanced driver assistance and autonomous systems. With an  enterprise-grade, long-term support model, the vehicle can maintain a fully-supported digital life cycle.

The platform can meet the changing automotive business needs by building expansive ecosystems and a common approach to standards through the power of the open source community.

Security

When looking for open source solutions for self-driving vehicles, security is paramount, and that’s particularly true in the automotive sector where safety is a key concern.

When using open source solutions, many have concerns about overall security. However, that’s starting to change.

Enterprise open source is increasingly seen as having many of the same positive attributes as proprietary software while also delivering on the benefits that come from the flexibility of open source licensing and the open source development model.

At Red Hat, we’re an industry leader in open source security, hardening software to make it safe for industries, such as automotive, where security and safety is paramount. Along these lines, Red Hat has set its sights on delivering the first continuously certified Linux® platform for vehicles.
Learn more about what’s new in automotive software.