Artificial Intelligence: The Future of Fully Autonomous Vehicle
The advancement of technology has never ceased to amaze humanity which influences the progress of civilization towards the future. It is considered as one of the major points that really brings significant impact to people’s lives, very well adopted and integrated with them. In ancient times, people utilized animals to support most of their activities such as agriculture, farming, etc. They did that because they realized that there is a limit of what a person can do with all those tough labors.
A millennium has passed without any significant breakthrough until the dawn of the industrial revolutions. The industrial revolution is the major threshold for civilization in entering and embracing the future, from agriculture and handicraft labor based economy to the utilization of machines manufacturing and industry which transform and increase productivity on a whole new level. So far, we are already at the fourth industrial revolution stage. In this stage, the whole industry elements like the physical, digital and biological are intersect and integrated within the whole end to end business processes. This technological breakthrough completely changes the competition landscape holistically and it really drives companies to stay on edge in order to maintain their competitive advantage strategically.
Through the technological advancement and existence of things like Artificial Intelligence (AI), Robotics, Internet of Things (IoT), Genetic Engineering, Quantum Computing and many more. It’s always interesting discussing all of these cutting-edge technologies and seeing how they are really helping in terms of business and better more change the way people live and it becomes part of norms.
Back in the days, we often watched sci-fi movies like star trek, star wars and many more. It gave us the visualization of how the future will look like and how the people live among the aliens. As for aliens, it still has not been solidly proven just yet. Among other things that was visualized there and perhaps one of the interesting ones is the self-driving vehicle or the popular term these days, “Autonomous Vehicle”. In this essay, the author will discuss more deeply about the Autonomous system as part of Artificial intelligence, what it is, how it works and how the future will look like with this technology being implemented in all types of vehicles by the whole world in every country and what challenges it presents.
Artificial Intelligence
Today AI is one of the trending topics in science, technology and Business. All big companies and emerging startups nowadays are investing a lot in Artificial Intelligence in order to gain competitive advantage.
Start with the ‘What’, what is Artificial Intelligence (AI)? Artificial Intelligence is the science and engineering of making machines and computers become intelligent by feeding them with lots of robust dataset. Through this process, computers and machines will conduct and experience a robust learning process which is now known as ‘machine learning’ and ‘deep learning’ to make them become even smarter and accurately solve problems. In general there are two types of Artificial Intelligence, mainly Artificial Narrow Intelligence (ANI) and Artificial General Intelligence (AGI). Artificial Narrow Intelligence is a supervised type of artificial intelligence or supervised learning. Supervised means that it can only perform a specific task that has been programmed into it. Like for example; YouTube, Netflix and Spotify, these applications are powered by Artificial Intelligence which is the supervised one. It can provide its users with movie, music or video recommendations that align with its users expectation based on its user’s history and preferences, this later known as SMART recommendation. So, Artificial Narrow Intelligence in conclusion is AI that was programmed to specifically solve ‘X’ problem. Next is Artificial General Intelligence (AGI), this type of AI is about making machines have the same intelligent capability and capacity as humans. The perfect depiction of Artificial General Intelligence in sci-fi movies would be just like in “Terminator” or “a Space Odyssey”. However it still has not yet to exist in today’s time and probably in the near future, as there are so many theories and anticipation about it but maybe it will emerge in the next 20 or 30 years in the future.
In Artificial Intelligence there are several subsets, like Machine Learning, Deep Learning, Natural Processing Language, Computer Vision, Speech Recognition and many more. In this case we will specifically discuss the application of Artificial Intelligence on Vehicle which today is known as Autonomous System.
Autonomous in AI
The Autonomous system in Artificial Intelligence is about how we make machines and computers to operate on their own without the needs of human guidance and minimum interactions. In the context of vehicles, having the capabilities to go fully autonomous can bring radical improvements, especially from the perspective of Efficiency and Safety. By having a fully autonomous vehicle, it can release the burden of drivers and actually support human’s productivity. For example, when you are in the need to do business travel from city A to City B and you have a lot of paperwork that needs to get done. Here is where it will benefitted the most, you can literally just set your designation point in the GPS system and turn on the autopilot feature, then get going with your paperwork while the car will self-drive you to your designation point. Another example, you just finished your work at the office, want to go home and you see that traffic jams are everywhere. Lets just say that if there is no traffic jam the ETA will be 45 mins to home and with traffic jam it can become like 2–3 hours on the road and that is really stressful. By using the autopilot features on your vehicle, you can just turn it on, hands-off the steering wheel and enjoy the sights while reading and have a cup of coffee.
How does it Work
So, how does Autonomous Vehicle actually work? As for this explanation I will be referring to how Tesla’s autopilot works. Based on the explanation given by Andrej Karpathy, the Senior Director of Artificial Intelligence and Autopilot Vision, that the keys of building the autonomous system in Tesla are machine learning and image recognition. In order to get Tesla’s vehicle to become fully autonomous, it needs to be fed with lots of datasets of images and labels so that machine learning will be able to learn from this data and get smarter. The dataset here must be large, varied and real (not from a simulator). So, here tesla built and applied its neural network learning and architecture software to get the vehicle become autonomous. Why? Because the way human vision and computer vision works are very different. Inside the human brain there exists millions of neurons that help humans recognize images and objects. The mechanism of human’s visual recognition is like, firstly the light enters the eye’s retina and from there continues to the visual cortex of the brain and then on to the millions of neural networks inside the brain and translates what those images and objects. Meanwhile computer vision works very differently than humans, it needs to be fed with tons of images, labels and scenarios so that the machine will be able to detect with high precision and get the car to drive naturally. The way of Tesla getting dataset of images is by placing 8 cameras in every angle of the car and those images will be uploaded to the cloud and have the machine learning do the process then update it back to the software. Sometimes machine learning is unable to do the recognition accurately because there are tons of unique visuals as the car drives along the road that seems a bit blurry. So, in order to make it work sometimes it still needs the help from the engineer to put labels on it to make the machine know that the roads are sometimes not always straight like the highways but it could sometimes be slippery or shadowed and many more.
Another thing that is attached to the Tesla’s car is called Lidar. Lidar stands for Light Detection and Ranging, it is a remote sensing technology that is used to determine range and surroundings. Through this lidar, a Tesla car is able to control its speed, distance and maneuver safely as it adjusts to its surroundings and architecture in 3D sense. Another approach that Tesla uses to gather this dataset is through what it called “Fleet Learning”. Fleet learning is basically allowing tesla’s AI and Autopilot engineer to export or request dataset from tesla’s fleet from around the world. So far there are about 1 million tesla cars scattered across US, UK and Asia and each of this country has its unique terrain and road dynamics which everything recorded by tesla’s vehicle. By obtaining these varied images and architecture, it gives the machine learning more learning materials which later will be updated into the software integrated system, so that no matter where you are if tesla is already there then it can drive fully autonomous without a hitch.
Also, whenever a Tesla car is being used manually, the computer vision and lidar keeps on detecting and uploading dataset to the system for continuous learning in the iteration process. Here the plus point is, the tesla autopilot system will be able to learn and recognize unique road patterns and driving behavior, so that next time when it switches into autopilot mode it already knows its ways.
Recently Tesla has upgraded its software and now the Tesla car has the capability of doing “Smart-summon”. What is smart-summon? Smart-summon is that the car is able to fetch and pick you up from its parking spot. Imagine you are at some buildings and it was raining with thunder and lightning sparks everywhere and you didn’t bring your umbrella with you. In this situation, you can actually summon your tesla to pick you up. The very first time I saw this smart-summon thing was when I watched one of Batman movies that starred Michael Keaton. In that movie, the batman actually summoned his bat mobile by speaking through its car keys. So here we can say that, when it comes to technology, really the imagination is our limit.
The Future Outlook with Fully Autonomous Vehicle
If we take a look at the future prospect of autonomous vehicles globally, we can say that it’s so prospective and will be profitable in the future. Statistically speaking, the market value of autonomous vehicles in 2019 is 21.4 billion dollars and by 2030 it was predicted that the numbers will be around 60 billion dollars. it will keep rising as the technology keeps on improving with all the other auto companies following Tesla’s footsteps. Recently there is news that the aviation industry has already started to adopt fully autonomous aircraft. For example, Airbus has already successfully tested its Autonomous Taxi — Take off — Landing (ATTOL) software.
Autonomous flight — Autonomous & Connected — Airbus
Airbus’s Aircraft already possess the capability to drive around the airstrip, take-off, fly and land automatically without the need for pilot’s intervention. In order for Airbus to make this autonomous software work safely, it took a total of 500 test flights and more than 80% of those test flights were to gather dataset, specifically images and surroundings just like Tesla’s autonomous vehicle.
As good as it looks, we cannot ignore that for the whole world to adopt fully autonomous vehicles in daily activity will present many challenges. We can use Indonesia for example. Indonesia is an archipelago country with lots of islands and cities and not all of them are already completely exposed to the internet and supportive infrastructures like good roads, civilized drivers and many more. I have no doubt that if one of that Tesla car driving around in Jakarta will present no problems, but what about if it’s driving in area like Villages or Remote area where the road lanes are not very clear and most of them doesn’t have any lanes at all or if there is a middle-aged lady driving around its motorcycle and she wants to go right but deploying the left lamp signs. This could lead to an accident and makes the autopilot software of an autonomous vehicle becomes stupid because of unable to drive properly and making the right predictions.
In conclusion, the advancement of autonomous systems in Artificial Intelligence is still far cry from what we expect them to be, but as soon as the transition of the whole world to Autonomous Electric Vehicle and machine learning to process all of the robust data from it, then it can speed up and allowing all autonomous vehicle to drive anywhere in the world.
