Tesla and Lidar

Using LiDAR to get one step closer to fully automated vehicles and how Tesla is trying to accomplish the same goal but going different paths.  LiDAR, radar, and sonar all sound so similar they got to have something in common right???? That was my first thought when I first heard about LiDAR.🦻 After a quick google search💻, I found out that LiDAR, radar, and sonar are actually very similar to one another. All 3 of these have the same goal: creating an accurate 3D map of their environments. 🗺 The difference between the 3 of these is how they create that 3D map. Radar uses radar waves, sonar uses sonar waves and LiDAR uses light. Hold up! Hold up! Hold up! I forgot to introduce you guys to LiDAR. Light Detection And Ranging are what LiDAR stands for and it shoots out thousands and thousands of lasers within seconds to create that 3D map of its environment. LiDAR is used to measure the elevation of things such as trees, buildings, cars, people, and nearly anything by shooting out those lasers. But wait, aren’t lasers used for other things such as scanning barcodes or even cooler playing laser tag😎.  Now, how exactly does this whole LiDAR system measures those trees and buildings and everything else??? Well, for LiDAR to work they need to collect data. There are 3 ways to collect the required data. From the air From the ground From outer space We are not going to focus deeply on how LiDAR gets its data from the air or outer space, but I would leave a couple of links at the end of the article if anyone is interested. On the other hand, let’s dive into how LiDAR works on cars and how there are already a couple of companies using LiDAR to help accomplish autonomous vehicles. Just a reminder, there is no company on this planet that has reached fully autonomous vehicles, but there is a company that is close 👀👀👀. First thing first, the way or process that LiDAR uses to create that 3D map of its surroundings is called Machine Learning. The whole Machine Learning is an extremely complicated process, but it is also extremely important. Machine Learning is basically a university for robots where they are continuously tested and trying to find a robot that is 99.99999% correct. So what’s happening is for example the system wants to know the difference between cats and dogs. A bunch of robots is then created and sent to the university. At the university, they’re going to show and test the robots on pre-labelled pictures of cats and dogs.  Source: CGP Grey In the beginning, these robots are really dumb and they truly don’t do good on the tests. Just like a student who first starts university and doesn’t receive those 90s and 100s they received in high school.   Source: CGP Grey But just like in real life, there is always that odd one out who gets a pretty decent mark. Same thing in machine learning, where there is that one robot who magnificently for example gets a 70%. What will happen next is that they take the robot that got the 70% and creates another bunch of robots, but this time they based it on the 70% robot. All the others robots who did worse will be thrown away. R.I.P 😢.   Source: CGP Grey Then they do another cat vs dog test and find another robot that did better than 70%. Machine Learning repeats this cycle millions and millions of times and each time they find a better robot. The simple formula of machine learning is more data = more tests = better robots.   Source: CGP Grey This cycle continues until they find a robot who becomes an expert at sorting pictures of cats and dogs (even better than a human🤯). Then the company crowns (not literally) that robot and the robot has graduated from the university. Trying to find the robot that receives that 99.99999999% is what companies call chasing the nines. It is not enough to be 99% accurate. It NEEDS to be 99.999999% accurate.   Source: CGP Grey This process of machine learning is EXACTLY how computers are going to do self-driving. The tests are going to be concentrated on different tiny pieces of driving and just run through the machine learning systems of finding the bot that’s closest to 100% and do that for every piece of driving. Now that’s machine learning in a nutshell and it is how LiDAR collects its data in order to make that 3D map of its surrounding. It’s finally time for the FUN PART: Waymo is a software company that creates a self-driving system using LiDAR and attaches the system to a day-to-day car. Waymo has modified the Chrysler Pacifica Hybrid minivans to be autonomous.  YES, you read that right, autonomous, meaning there is NO ONE driving the car. Cool eh? 😎😎😎. Waymo is the company that is closest to accomplishing fully autonomous vehicles, but they are not there yet. The MAIN reason that Waymo has a fleet of autonomous vehicles is because of (you guessed it) their one-of-a-kind LiDAR systems. FASTEN YOUR SEATBELTS BECAUSE WERE GOING FULL SPEED ON THE BENEFITS OF LiDAR ON AUTONOMOUS VEHICLES. One of the best features of LiDAR is able to create an EXACT 3D map of its environment. And when I say EXACT, I mean a whole 360 degrees map of its environment including the smallest of details. Radars and camera sensors aren’t able to get 360 degrees of vision of their environment and the 3D map is not as detailed as the map created by LiDAR. Look at those details 😍 Since the 3D map is super accurate, it helps with the whole self-driving since the car knows where everything is. Knowing the surroundings will limit the number of accidents caused by autonomous vehicles. 2. LiDAR is able to work both during the day and during the night. These are the only advantages of LiDAR, but it is all about the quality, not the quantity. IF LiDAR IS EXCEPTIONAL IN SELF-DRIVING CAR, WHY DOESN’T TESLA USE IT THEN?????? Well, it’s because Elon Musk hates LiDAR a lot. Like a lot. He stated this about LiDAR:   Literally, this guy got beef with LiDAR. Like DAWG what did LiDAR do to you😂. Elon Musk also stated that he can achieve autonomous vehicles without using LiDAR by 2020 Cough* cough*, Elon we’re still waiting on that to happen 😁 There are also other reasons why Elon hates LiDAR and most of the reasons are well…. reasonable. It’s EXPENSIVE VERY EXPENSIVE 🤑🤑🤑 It is going to costs $10,000 extra to add a LiDAR system to Tesla vehicles and I honestly don’t think that Elon wants to increase the price of Tesla because no one will buy it. 2. FRAGILE EQUIPMENT This equipment is extremely FRAGILE The way that LiDAR works, physically is that mirrors are spinning extremely fast which can cause them to wear or tear and break down over time. Trust me if even the smallest piece wear or tears or worse breaks down, the WHOLE LiDAR system is going to stop working. Memorize this simple equation: Little damage = no self-driving  3. NOT NECESSARY (According to Elon) The whole Tesla auto-pilot system is made by computer vision which is a type of machine learning where the computer learns how to pick out images Elon Musk is betting his entire paycheck on computer vision in order to achieve autonomous vehicles A fair point made by Tesla is that the road system is made for human drivers with regular vision. It wasn’t made for lasers to shoot out of our eyes and create that 3D map of the surroundings. We actually create a 3D map of our surroundings, but it’s just with our bare eyes 👀. The way that Tesla creates their vehicles is by using reverse engineering. They reverse engineer the human visual system by using machine learning except using 2 cameras (our 2 eyes), they use 8 cameras to help see in all directions.  Tesla computer vision 4. IT’S UGLY LOL 😂 This hasn’t been publically spoken about by Elon or any other members of Tesla, but common we all know that Tesla looks cooler than Waymo. Could y’all imagine a Tesla with LiDAR on it because I can’t?  LOOK AT THIS BEAUTY 😍😍😍😍😍 The LiDAR system is tremendously useful in the autonomous vehicles industry and all the benefits have been demonstrated by Waymo, but I don’t see Elon Musk and Tesla using LiDAR in the near future. I’m not here to say who is going to win the race to fully autonomous vehicles between Waymo and Tesla, but I'm here to say that there can be 2 ways that fully autonomous vehicles can be achieved. Both Tesla and Waymo can accomplish this objective in completely different paths which is totally OK! #tech #tesla #Lidar