| 6 Nov 2017
Driverless car, or the self-driving car, is the new buzzword in technology. Through this blog series, we will try to unfold the mystery around driverless cars by tapping into the mind of Mahbubul Alam, CTO & CMO – Movimento Group, a Delphi Company. He is an avid technologist and entrepreneur, engaged in the development of the Software Defined Car™ in association with ValueLabs.
Q. Why driverless cars? What is the need?
A. There are three main benefits provided by driverless cars:
Q. How are self-driving systems different from driving-assistance systems?
A. SAE has classified automated driving into six levels of automation, ranging from zero automated features to fully autonomous vehicles.
Today, though humans are still responsible for controlling the vehicle at short notice, it can do many operations by itself – such as brake, accelerate, follow the road and follow / avoid other cars. Ultimately, we are heading towards a complete self-driving system, where the vehicle can go from point A to B without any human interference.
Q. How does a driverless car see the road?
A: A driverless car sees and gauges the road through multiple means. Any camera can see things. It is not just about seeing objects, it is about sensing, detecting, classifying and recognizing the objects and then actuating accordingly that makes driverless cars so exciting and interesting. It has a computer vision that is enabled through smart cameras through which it recognizes objects. It also uses short-range and long-range LIDAR, RADAR, and proximity sensors to create images of the road and even objects on and off the road. The fusion of data from all the different types of sensors in the vehicle along with deep learning algorithms (family of artificial intelligence) enables decisions for autonomous vehicles, also known as driverless cars.
Q. How would you describe the experience of having a driverless car?
A: Imagine a situation where you are sitting next to a driver who is making you feel uncomfortable with the driving. Scary, isn’t it? On the other hand, imagine that your car is navigating smoothly through traffic and you are able to utilize the extra hours of commute, which usually get wasted, to become more productive.
With driverless cars driving you from Point A to B, you would be able to do more in the same time, for instance, spend more time with family, indulge in leisure activities or just get more work done. In short, it would be like a personalized high-speed train experience, where you do not control the vehicle directly but you get to your destination seamlessly with facilities like Wi-Fi, movies, time with family members, etc.
There will also come a time when people will not own cars or even driving licenses! You will just need a mobile app to call in a driverless car when you need it.
Q. How do you visualize a driverless car tackling daily challenges on the roads?
A: In my opinion, with automated driving, there will come a time where remote human operation centers will be responsible for driving safely and efficiently in different regions. For example, take the city of Cupertino, California. There will be a couple of back-end operators who would be responsible for Cupertino Automated Driving Safety. The cars would be designed to enable the capability of estimating and predicting outcomes or issues. In situations where the vehicle is not being able to understand what to do, the vehicle would simply pull over.
Automated vehicles do not have the capability of making human-like decisions or judgements. Such scenarios will be sent through the network to the cloud where the operators can see the live feed and make decisions only for that specific moment, telling autonomous vehicles that they can make a situational maneuver. The industry will not only have automated drive but will also have sets of operators managing the back-end operations.
If we look at the AI family, there is machine learning but it is still not sufficient for what automated driving needs. It needs advanced deep learning, a sort of deep dive into behavioral and situational learning, and goes beyond machine learning. This is what is required to mimic a human brain to make decisions. There will always be self-learning, but just that alone is not enough, the vehicle needs to learn the experiences of other vehicles as well. All the experiences gained by the vehicle should go into the cloud where they are recorded, stored and distributed to all other vehicles through updated algorithms and the recognition of new objects and situational data.
In the next post in this series, read more about the evolution of driverless cars, the challenges that the industry is facing, and more…
The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.