Can A New Algorithm Improve The Safety Of Self-Driving Cars?
It was the May of 2018, the sudden demise of Elaine Herzberg had created a buzz, emanated by many fans and opposers of self-driving cars around the globe.
The vehicle's automatic systems failed to identify Ms. Herzberg and her bicycle as an imminent collision danger in the way they were supposed to, the NTSB found. As a result, Uber's ability to test self-driving cars on the state's public roads of Arizona was suspended.
Until now we had heard how most car accidents happen because of human errors and that self-driving cars would change that. But this incident put a big question mark on the safety of autonomous vehicles. As self-driving car companies became notably more cautious in their trials, a question took rounds among people- can a new algorithm improve the safety of self-driving cars?
Fast-forwarding to 2021, while we still have miles to go before we can completely trust self-driving cars eyes-closed, many companies have come a long way in developing new algorithms to make them safer by the day.
In a patent published last year, Waymo, a subsidiary of Alphabet Inc. described a new safety feature to keep its passengers safe and keep their apprehension at bay.
The invention is both literally and figuratively a “safety-net” positioned on the hood of the car.
In the event of an impact, this net will pop out of the hood of your car stop the stray shrapnels from smashing the windshield and entering the car. In case the car inevitably hits a pedestrian, the net will also save the pedestrian from being injured by the bumper of the car.
The patent also mentions a shock-absorbing mechanism, so most of the impact from the collision will be absorbed by the device attached and you’ll come out of the collision with minimum damage to your car.
Another recent development by Wipro is the Intersection Management System. What’s that now?
Well, apart from collecting the information from its surroundings a self-driving car usually relies on traffic signals to maneuver its way around to function well on the road. But what happens if someday the signal stops working?
Worry not, Wipro has come up with a sensor that will understand the signs of Traffic Police as part of taking in and processing the surroundings. Their recently filed patent describes that the sensors will be able to detect hand gestures by traffic police.
But what happens if the sensors of the car fail altogether or some of them start malfunctioning?
Zoox, an Amazon company that is working on driverless car, has got that covered for you long before!
The AI system in the car will automatically detect the malfunctioning component and guide the car to compensate for it using alternate sensors. The camera in the front of the car isn’t working? It will ask the car to switch to the LIDAR sensor in the front.
The patent published in 2017 by Zoox further talks about a remote computer that will monitor the entire fleets of driverless cabs and look for such help requests. And while the car compensates for a malfunctioning camera, the remote computer will find the safest route to the destination and then to the nearest autonomous vehicle service station. The same way a human is guided when he is lost.
Companies are working exhaustively to make an already revolutionary idea of self-driven cars even more outstanding, implementing methods and techniques that can make them the safest option.
During the last few years, several automobile giants have announced their plans to launch fully autonomous cars in the next 5-10 years. And, conceivably, with the help of these innovative ideas, the prospect of self-driving cars running safely on our roads will be a close reality.
