It is 2020. Where are our self-driving cars?

 

A few years ago, while the discussion was about self-driving cars hitting roads, we in India were wondering how can a car with AI drive on Indian roads. We don’t have proper markings and demarcations; each road is different from the other. Our system of getting a driving license is even flawed and corrupt, because of which it is easier for people, who do not know how to drive properly, to get a license which results in accidents and bad driving on the roads. We were so engrossed in those dreams then, without realising, that making a car driven by AI will not be easy. It is not our fault though. Some leading automobiles manufacturers promised to deliver them by 2020. Tesla, said by 2018, which the later changed because they couldn’t. The article by ‘Future Perfect’ pens down the current situation of self-driving cars. It describes all the aspects that are connected to AI- from history, to present mentioning the safety and the need as well. It has 9 sub-topics that cover each aspect. I picked up three very important topics from the articles that I will give you an insight into- How do these cars work, the safety issues, the current scenario and the future.

How do AI cars work?

Self-driving cars work by anticipating the road and obstacles ahead, using high-resolution cameras, light detection and ranging sensors. It will have internet-enabled maps that can drive from one place to another. Sounds simple right? It actually isn’t. Any AI depends on a lot of data. We don’t have recorded data for any situation that might come across while using an AI-based car.  Even in our day-to-day situations, our human intelligence, which is evolving in nature, fails to work in a certain unseen situation which we may face that results in an accident. There are also gestures and understanding that takes place in the last minute between humans, which AI can’t replicate. Driving is one of the most complicated activities that is performed and to rely on a software which can only work in the situations it had been shown before, is a big risk to take. Even the surveys conducted show the unwillingness of passengers to travel in a car like that.

The safety issues

No matter how much engineering is put into it, our technologies have never been completely safe from accidents. In March 2018, a self-driving Uber noticed a woman, who was not walking on a zebra crossing and didn’t have data to understand the situation. It used brakes just a quarter of a second before hitting her hard. Even though the report said all the sensors were not at use when it took place, it showed us a big flaw that can never be covered- we can’t have data for potential accidents because those are different from each other. Tesla also had such an incident in the past where its autopilot rammed into a divider killing the passenger. Safety is the primary issue why the companies are not rolling out the cars on road, despite having run it in a restricted environment for a long time now. As a driver myself, who is sceptical of overuse of technology, I would not sit in a car driven by AI, which can never judge by itself because it lacks the ‘gut feeling’.

The current scenario and the future

Despite understanding the fact that there is a safety issue and that people will travel more (because the driving strain is not attached to it) which will cause more emissions, we still want to experience it once. Tesla, despite claiming to make a self-driving car soon, has only been able to produce one that has an autopilot mode, which efficiently works on Highway and needs human touch every 10 seconds. Waymo is working on it for a long time, even on a steering less car, but nobody knows when will it hit the roads. Even though multiple companies are working on it for a decade, no one wants to take the risk of launching a car until it is a full-proof one. An issue or problem can cause severe damages to the reputation, apart from lots of financial loss. But we might get to see one in this coming decade, because the first company to engineer this marvel, will be able to be the leader while the other catch up to it. It will be open to fortune for the company as many petrol heads and people with big pockets want it already.

Conclusion

The article was interesting to me and kept me glued to it. It has enough information to make you understand about AI even if you do not know anything about it like I did not. The author has added a bit of personalisation. I like the detailed report of the accident that took place with Uber’s self-driving cars. I would suggest you go through the entire article as it was very detailed yet not complicated to read and understand.  It is a 10-minute read at most.

x

x