Is AI the Answer to Achieving Complete Autonomous Driving?

Artificial Intelligence (AI) has become an increasingly common term in discussions about cutting-edge technology, often overshadowing phrases like “Driverless,” “Autonomous vehicles,” “SDC” (Self-Driving Cars), “Smart cars,” and “Connected cars.” There’s even a specific domain extension, .ai, dedicated to it.

For instance, a search on LinkedIn reveals hundreds, if not thousands of professionals engaged in AI and its subsets, such as deep learning and machine learning.

Why is AI a Growing Field?

The surge in AI activity is largely driven by the significant progress in autonomous driving technology, which remains a lucrative segment in the automotive industry. In my opinion, without AI, true autonomy will be unattainable, leading to safer, more efficient self-driving vehicles.

Vehicles require much more than expensive hardware—Lidar, cameras, sensors, and servo motors— to operate safely and independently. They need a level of intelligence that has yet to be fully achieved.

An Everyday Example
Consider a typical scenario involving a navigation system used in a conventional vehicle. When the navigation instructs a driver to turn right, they might find that the turn leads into a private driveway. As human drivers, we instinctively recognize this as a minor error and adjust accordingly, realizing that the correct turn is just ahead. This quick correction is a demonstration of human intelligence.

Even though this task seems simple, replicating such swift decision-making in machines requires advanced Artificial Intelligence.

Understanding AI

A clear explanation of Artificial Intelligence can be found in this short video on the BBC website.

Currently, cars do not possess the type of intelligence needed to handle even basic driving tasks autonomously. For instance, can a driverless vehicle understand hand signals from a traffic officer? Can it navigate safely over broken glass without stopping or swerving dangerously? Or, will it recognize when a child might suddenly run into the street, something a human driver would instinctively anticipate?

These situations, along with countless similar scenarios, require immediate human judgement to ensure safe navigation. The term infinite is crucial here, underscoring that machines cannot be pre-programmed to handle all possible incidents.

The ability to think and respond like a human is essential for effectively managing the complexities of everyday road traffic (this may even mean that AI could occasionally make mistakes, similar to humans, though that’s a topic for further exploration).

The Future of AI

I believe that until AI reaches a level where it can serve as a reliable observer and operator, we will primarily see semi-autonomous vehicles in the coming years. Some suggest that a combination of semi-autonomous and fully autonomous vehicles could emerge by 2020, especially in controlled environments like the Tokyo Olympics. I prefer the outlook of 2025, which is not far off, especially given the substantial resources being invested in AI and deep learning.