Our Future Reality or the End of the World
An intro to Self-Driving Cars
You know those old movies where they are like “robots are taking over the world” and that being the cause of the end of our existence?
Well, to be honest with you, it can happen but it’s really unlikely. But these robots were the past, and who knows how they were built. Those robots couldn’t really just use a human brain to think and actually do things; they need some sort of brain that works with technology. So they are probably powered by artificial intelligence.
I don’t really know if they were powered by AI, but I know something: autonomous vehicles. You probably may have heard about self-driving cars already, but you may not know exactly what makes them autonomous.
To start us off, if you don’t know, autonomous means for something to control itself, and a vehicle transports things like people or goods. Even though it may sound simple, there are certain aspects for the car to drive by itself.
So you may be thinking, “You literally just told us that those movies that have robots taking over the world using AI, and you are saying that we should trust these cars that can work by themselves? Well, first of all, what is the maximum thing these cars can do to harm humans? Like are they gonna just drive over us?
Right now, there are 1.3 million people that are killed each year in car accidents. If we can try and improve this rate, why not?
To actually get a car to drive on its own needs to be programmed even better than a human. This means it has to have a better reaction time and cancel out all the faults that we human drivers make on accident. The cars also have to be programmed to handle every situation like constructions, pedestrians, bikes, potholes, and vehicles driving around them. To do this, some technologies can be programmed to help the car function.
🚘 5 LEVELS OF AUTONOMY
As I said before, autonomous means for something to independently control itself. If you apply this to self-driving cars, there are 5 levels of automation:
(1) No Automation/Driver Assistance,
(2) Partial Automation,
(3) Conditional Automation,
(4) High Automation
(5) Full Automation.
The first level is No Automation. It’s as simple as it sounds because this is a car that cannot do anything without a driver. AT this point, the car is like a newborn baby, helpless and cute (well, not the cute part). This baby cannot walk or basically do anything alone since it needs a parent to help it survive or, in this case, a driver to drive.
The second level is Partial Automation. This means that the car can move on its own, but the driver should be ready to take over at any time. For this level, the vehicle can recognize its surroundings but is not fully prepared for an emergency. At this point, the car is like a baby that knows how to crawl but needs some help standing up to try to take some steps.
The third level of autonomy is Conditional Automation. This level is very similar to partial automation, but there is a very low chance the driver has to take over. Mostly the driver will take over in unknown situations, like if a vehicle nearby decided to change lanes without an indicator. This level is similar to when a baby is trying to walk but occasionally needs a parent’s hand. The baby can walk by themself, but they need some kind of support.
The fourth level of automation is High Automation. This means the car can drive by itself, but the driver still needs to sit in the front seat and rarely needs to take over if something unexpected happens. This stage is like when a baby knows how to walk by themselves but occasionally tripping a bit but is comfortably walking without any help.
The last level of autonomy is Full Automation. At this stage is where the self-driving vehicle is 100% independent; no help is needed. In this case, the driver’s seat can become a passenger, and the car would do all the work. This stage is like a baby, walking all by themselves just like an older kid and never needing human assistance to walk.
There are many advancements in self-driving companies, as many companies in this field are improving their self-driving car models. Even though there is a range of companies working on this, they aren’t all at the same stage yet.
Let’s start with level one; No Automation. There are too many companies that are not autonomous at all. Some of those companies would be like regular car companies like Volkswagen, BMW and Toyota. Although these companies sell cars, they partner with companies like Mobileye, a company specializing in driver assistance systems. By partnering with these companies, regular car companies are branching out to newer ideas.
The next level is, of course, level two for Partial Automation. You have probably heard of the famous eclectic car company tesla. They also make self-driving cars and are at stage two currently. You may be saying why such a big company is behind many other companies, well they have a good reason. They are looking for ways that will be the next big thing and not just going with what we have right now. They are trying to look at all the faults in the technology used for autonomous vehicles and improve them. (I will discuss later on what exactly they are doing for this)
The third level is Conditional automation, and that’s where Waymo is at. Waymo is not too popular since its cars have not been open to the public but are still being tested more. This company is one of the leaders making self-driving cars but is still aiming to advance to higher levels.
The fourth level, High Autonomy, is where very few companies are at and for different reasons. One leading company is Cruise, a rude hauling service. They are at a bit of the 5th level and are still testing and not seen on the roads yet. There is a lot of working upon, troubleshooting and improving through the whole process to guarantee the passenger. This way, people won’t be reluctant to ride in a car that can drive itself.
The fifth level of Fully autonomous has yet to be unlocked. Many companies are still undergoing testing, but none of them can actually be seen in real life. Companies are working to create sustainable and cost-efficient ways to make these cars. Someday, the new normal would be having a car and just relaxing to drive it.
💡Key Takeaways:
There are 5 levels of autonomy (in order of no to full autonomy
1. No Automation — Driver needs to help
2. Partial Automation — Small steps by itself
3. Conditional Automation — More independence, occasional help
4. High Automation — Rare need for help
5. Full automation — No help at all
🧠 THE BRAIN
Now that you know about levels of autonomy, there has to be something that actually powers the car and helps it think, just like a human. Almost like a human and how it thinks, it needs some sort of processing system like our brains.
There are five main physical features to autonomous vehicles which helps them, recognize the process and actually put it into action:
- Computer Vision
- Sensor Fusion
- Localization
- Path Planning
- Control
First, iis Computer Vision. Computer vision is almost like our eyes, but it’s constantly taking pictures. This step basically takes the digital image of the view around the car. Depending on what strategy you want to use, you can use different technologies. Tesla uses cameras around the vehicle and takes many pictures which are digitized.
Basically, this step is when we see an image of a friend. Without actually distinguishing it, we use the clues to identify what is happening. We first distinguish it from the background and the person. Then we use our previous information that we know and clues to indefinitely who that person is. This is when the pictures are processed from the cameras and put through an Ai system recognizing the shapes to identify the objects.
This leads us to Sensor Fusion. This is when you use sensors to identify, spot and figure out everything you need to know about the surroundings. Computer vision or using cameras is one way of doing the seeing process because there is another famous technology; LidAR. LiDAR is a bit more common and has its own advantages and disadvantages. LiDar is basically a rotating sensor (placed on top of the car), which sends thousands of lasers every second to different objects. This then Identifies where they are, the rough shape of the object and how far it actually is.
You may be asking how Tesla figures out the distances between objects. Well, they use something called Radar, which companies using LiDAR also do for different purposes. Radar can be used to find distances and sense its surroundings through the different weathers as well. Waymo is a company that uses a fusion of LiDAR.
Using either cameras or LidAR, they both go through an AI program that identifies the object with the information it needs to avoid collisions. This applies to moving and still objects no matter where they are in 60 metres of distance.
If you want more information on this topic, I have published an article focusing on computer vision and Lidar, which goes more in-depth: https://manaallakhani.medium.com/inside-a-car-with-its-own-mind-960d792c3b42.
The third step is Localization which is when the vehicle uses different landmarks from the previous knowledge to navigate where it is. This step helps the vehicle identify if it’s going in the right place or not. They provide this information by looking at the pictures taken by either the LiDAR or cameras. One thing that really helps in this process that only cameras specialize in is reading specific road signs like street signs to navigate better.
The second last step is Path Planning is when from the information received, like the distance from objects and Localization, it identifies where it has the space to drive. Without the crucial information, the car can go onto the sidewalk, but it will go straight to recognize that there is a sidewalk right beside it.
Lastly, we have Control. Control is very simple because this is the combination and the final outcome of everything. This where it uses every bit of information it has and lets the car navigate. This step enables you to ride in a car by itself, just like it has a human mind but more intelligent.
As you can see, we cannot just plop a pair of eyes and brain into the car because, of course, that won’t do anything, but it needs a pair of eyes with a brain metaphorically. The eyes would be the cameras and sensors, and the brain would be the software that recognizes everything it needs to. AS we advance to the higher levels of autonomy, we might start seeing people sitting in cars without having to drive at all.
Hopefully, this gave you a good surface-level understanding of the levels of autonomy and the processing systems. If you enjoyed this article, give it a round of applause.
Make sure to check out some of my other articles.
:)
