Self-driving cars: How close are they to hitting the road?

Autonomous vehicles (AVs) are the future of mobility, so the question on everyone’s mind is: When will they arrive? It’s an important question, and one that OEMs and startups alike hope to answer in the near future.

Because driverless road testing is limited to selected cities and test sites, most consumers haven’t experienced the technology firsthand yet. But it’s coming—and in time, self-driving cars will bring many benefits that will completely change the way we travel. When the car drives itself, every driver actually becomes a passenger. You can sleep on the way, relax and watch a movie, or answer emails on the way to work. Over time, as consumers think about how to spend time in the car, the interior environment may look more like an office or living room.

However, self-driving cars won’t arrive overnight. While the automotive industry has made rapid progress in autonomous driving, the reality is that we are still only in the development phase. Some suggestions mention that we won’t see self-driving cars on the road until 2034 at the earliest .

To embrace the next era of mobility, automakers need to consider many factors, such as safety, laws and regulations, infrastructure, consumer acceptance, data networks, ethical issues, etc. Therefore, it is very important to properly address all of these factors before self-driving cars hit the road.

As a world-renowned semiconductor IP manufacturer, Imagination Technologies has been in the automotive field for more than 15 years, providing high-performance solutions and high-quality support services for a variety of automotive electronics. At the same time, we have maintained a long-term focus on self-driving cars and launched graphics processing and artificial intelligence products that support related applications. Let us further analyze the above-mentioned factors related to self-driving for you.

Safety first

OEMs have always been committed to safety, and this is especially important for self-driving cars, as they will eventually move away from pedals and steering wheels in favor of an array of sensors and automated features. As cars gain more autonomy, they will need the ability to understand the road and all its complexities to ensure safe operation.

This is one of the main reasons why self-driving cars are taking longer to develop than expected. In traditional cars, humans come pre-equipped with a vision system (eyes) that is connected to the most advanced computer known to man (brain). The whole process is so seamless that most people don’t even consider what challenges they might face once they hit the road.

But a self-driving car must consider all that and more. It needs to be able to handle the coldest, snowiest day in New York just as well as it can navigate the bright, blinding sun in Arizona. Self-driving cars need to be as courteous as human drivers, communicating their intentions to pedestrians and human-driven vehicles. They need to handle changing road conditions (mud, ice, cracked pavement, etc.) and may even need to determine when driving is too dangerous to proceed. They also need to be able to make split-second judgments. For example, is it a piece of trash that has blown across the road or a child crossing the street. The vehicle needs to see, analyze, and react just as a human would.

By spending a lot of time experimenting with all of these factors, self-driving cars will have a huge advantage when they are ready to go. Testing is already taking place in many places, including the American Center for Mobility in Ann Arbor, Michigan, which has an autonomous vehicle testing site. The goal is to allow consumers to travel with confidence, knowing that these vehicles have been rigorously tested and fully polished before they carry their first passengers.

Attract consumers

Consumer attitudes are often cited as a major barrier to the deployment of self-driving cars, but a recent survey shows that this has changed. SAE International, a global organization for automotive engineers, interviewed nearly 1,400 consumers who had ridden in a self-driving car . The results were eye-opening: 82% of consumers were "enthusiastic" about self-driving cars before taking a test ride. After experiencing the technology, nearly 10% of consumers expressed even higher enthusiasm.

More than three-quarters of participants (76%) believe that the experience of driving an autonomous vehicle will be similar to or better than driving a traditional car. More than a third of participants (37%) believe that autonomous vehicles will be safer than human-driven cars. This doesn’t mean that consumers are ready to give up their current cars tomorrow, but it is a sign that people’s views on mobility are changing. It also suggests that as consumers experience autonomous vehicles firsthand, their attitudes toward the technology will only improve.

Making cities smart

There are two elements to the realization of future smart infrastructure and smart cities. The first is that there are some challenges to achieve the goal. Self-driving cars require both vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) technologies, both of which are still in the early stages of development. The second is the need for ubiquitous connectivity - 5G, Wi-Fi, dedicated short-range communications (DSRC) or some other method - and a large number of sensors, which need to be installed in places such as street lights, traffic lights, stop signs, etc. One self-driving car startup has deployed its own self-driving car sensors to make the car better at driving in the city, which is a very large project.

The technology needed to widely deploy self-driving cars is still under development and won’t be ready overnight. Like any road construction project, this is a multi-year initiative that will take some time. But once it’s done, the benefits will be considerable. Rather than investing millions or even billions of dollars in road expansion projects, developing smart infrastructure may actually reduce expenses. Smart infrastructure could allow driverless cars to travel closer together and at the same speed, eliminating the need for wider roads or extra lanes.

Get it right

In short, self-driving cars are the most important machines automakers have ever developed, and there is too much at stake to rush them. Therefore, a slow and steady approach is the right approach.

Imagination can provide all-round support for you in developing autonomous driving applications. For example, our PowerVR GPU IP, among all the automotive GPU chips in the world, 40% are designed with Imagination's GPU IP, ranking first in the world; at the same time, in order to better support future smart cockpit applications, we recently launched the new IMG A series GPU, which can provide the best performance with a very low power budget for a longer running time.

Another example is our neural network accelerator (NNA) IP. The latest 3NX NNA can provide up to 160 TOPS of computing power in multi-core state, which can well meet the requirements of Level 4 and Level 5 autonomous driving. At the same time, NNA and GPU can form a good cooperation on the technical level, taking advantage of both the high computing power of NNA and the flexibility of GPU. The combination of the two has the best PPA (performance, power consumption and area) indicators, which can provide excellent heterogeneous computing solutions for autonomous driving applications.