While our contemporary understanding of AI is eschewing the stereotypes of Sci-Fi and mid-century dystopian dreams, the fact remains that the current environment of rapid technological innovation and collaboration is one that has been in the pipeline for many decades already. The groundwork was laid in the form of ideas which, at the time, sounded far too advanced to prove attainable in such a short span of time.
One example of an idea that spans two millennia is autonomous transportation. For many years, we’ve heard about automobile companies trying to figure out how to replace the human driver with a reactive, anticipatory machine intelligence – one that proves safer, even, than we do.
From cars that can communicate with each other, to urbanized cities without traffic lights – all will be made possible with self-driving cars – but will they ever be ubiquitous?
The Biggest Challenge
Put simply, removing human drivers means developing a highly sophisticated AI replacement. The car will need to have sensors that can interpret everything it will encounter on the road. Usually, the process begins by using annotated or labelled data sets; text and video annotation represent indispensable features, essentially enabling the car itself to identify and behave accordingly around traffic, pedestrians, and other dangers.
Any vehicle given autonomy on the roads must be harnessed by a technology capable of detecting, monitoring, tracking, and assisting with an endless list of potential variables – variables that are encountered on any given journey made by any given vehicle. A technology capable of retaining a high statistical advantage over human drivers is awe inspiring.
Factors such as weather, and other random incidents that may occur on the road add to the complexity of the process. Can you imagine how difficult it is to teach a self-driving car to distinguish between a pile of leaves and a flock of birds?
Still, leaves and birds may seem like child’s play when compared with the feat of preparing a machine to respond to the infinite nuances and varying behaviors of drivers on the road. Once again, this is only a small part of the battle – and a clear indication of why autonomous vehicles are not gearing up to replace human drivers overnight.
Ubiquity Can Be a Reality, But Could Take Longer than Expected
Autonomous driving models are already in the works for some car manufacturers, such as Ford. The good thing is that the technology supporting this innovative concept continues to develop at a fast speed, and with remarkable accuracy. For instance, computer processors are becoming more powerful, yet compact. Artificial intelligence is becoming more human-like in many aspects, further supporting the requirements of autonomous transport.
Moreover, communication technology – particularly with the launch of 5G – is another critical piece of the puzzle. Faster wireless internet speed along with cloud computing are essential capabilities of successful self-driving software.
Although you may not see many autonomous cars driving down your street within the next couple of years, the technologies intended for this application are already in use. New car models today have driver assistance systems or ADAs. Some common features of this system include alerting drivers to traffic lights, managing cruise control options, and supporting emergency braking systems. The introduction of these features in car models today will familiarize consumers with what to expect with full driving automation. In a way, it will prepare them to accept what’s likely to come within the next decade or decades.
The road to seeing self-driving cars on the road is a difficult one. However, car manufacturers love a challenge and aren’t backing down. With the support of the tech industry, especially the ever-growing AI and machine learning field, anything is possible.
