WEEKLY WHEATIES #2512
LIDAR vs Cameras, Tech Headlines, Interesting Maps
In this newsletter:
📝 Post: LIDAR vs Cameras: The Self-Driving Tech Battle You Need to Know About
🗞️ In Case You Missed It: Amazon Firesticks, Pebble Watch, AI Copyright, Google Pixle
😎 Pick of the Week: Interesting Maps
📦 Featured Product: Blueair Air Purifier
📝 LIDAR vs Cameras: The Self-Driving Tech Battle You Need to Know About
Last week, a famous YouTuber (who is also a former NASA scientist), Mark Rober, posted a video on his channel titled, Can You Fool A Self Driving Car? It received a lot of flack for getting some of it wrong, and a potential paid advertisement for the LIDAR (Light Detection and Ranging) system used - which I personally doubt, but either way - it doesn’t matter. There’s a bigger issue here than arguing if that’s the case. Which Mark said wasn’t in an interview with Philip DeFranco.
LIDAR uses lasers to 3D map out spaces in a virtual or electronic space. It can help pinpoint items to the exact location, help with measurements, and allow a computer to ‘see’ what humans see. In a sense, it may also be compared to the movie The Matrix. Alternatively, any video game you’ve played that includes buildings and spaces recreated from real life could have (in theory) been mapped via LIDAR.
For vehicles (whether electric or not), LIDAR is and can be used for features like ACC (Adaptive Cruise Control). By shooting lasers in front of the vehicle, the computer can determine how far away the next vehicle is. As the speeds change, so does the time for the lasers to reach back to the original vehicle, telling it if it’s okay to speed up or if it needs to slow down.
In comparison, Cameras are used for features like LKA (Lane Keep Assist) that allow the vehicle to stay in between the lines on the roadway. However, Tesla has started using cameras for their Tesla Vision in place of LIDAR. Arguably, as they should. This leads us to the next point of discussion - Full Self Driving.
Tesla has also started to popularize the terms Autopilot (AP) and Full Self Driving (FSD) to explain the notion of the cars being able to drive themselves. However, Autopilot only allows for ACC and LKA, and it requires the driver to pay attention to the road - or be ‘supervised’ at all times. On the other hand, FSD typically allows complete ‘driverless’ navigation, including stopping at stop signs and signals, making turns, entering and exiting freeways, parking, and much more. There are still many quirks here, but against common belief, the driver must be supervising the drive at all times.
As a Tesla owner myself, I’ve ‘driven’ to work from my home many times without having to interact with the car at any point. However, there are still types of roadways and scenarios where it’s not perfect. Undivided highways with a middle lane, roads with little or no striping, and avoiding potholes are just some examples. Sunlight can also affect how and if the cameras are able to see, so there are potential issues with vision there. One potential issue I see with cameras is they only use one in the center of the windshield rather than two on the edges (to produce binocular sight). I’d bet this would help with distance control over cameras, but there are also people way smarter than me working on this, so…
You also can’t take away what FSD is actually capable of, though. A simple search on YouTube with the terms “Tesla FSD saved me from…” yields tons of results with real-world examples. I also realize there are examples where FSD does not and did not work well. But if we crunch the numbers, it appears there are fewer crashes overall in comparison when drivers are utilizing FSD, and probably even less when the driver is ready and does take over.
A friend and colleague of mine researched SAE Level 2 Advanced Driver Assistance Systems (L2ADAS) for his dissertation and ultimately shared how “Your Car Still Needs You in 2024.” As seen with the articles and rebuttal videos coming out about Mark Rober’s video, there is a lot of misunderstanding in terms and abilities of these ever-improving systems.
These systems shouldn’t be used as a game when trying to decide if it will react correctly, nor should the driver ever even consider sleeping while FSD (or a similar system) is active. If you have a vehicle with this option and use it, simply take over if you’re worried it may not react the way you would in any scenario, just as you were driving a non-equipped vehicle.
So where does that leave us? Well, after noting potential issues with the all-camera systems, what about issues with LIDAR? To be blunt, LIDAR cannot read road signs (speed limits/stop signs/etc) and signals, nor can it always make out what a given shape may be - especially if it’s stationary. The lasers used here also have issues with water (or rain in this case), and they also can be disabled as simply as the cameras can just by covering them up or placing something in front of them. LIDAR systems are more expensive and require more processing power (which is also more expensive) in comparison to an all-camera system.
Currently, neither system is perfectly capable of independently functioning properly to get us to the fully automated Level 5 I described back in Weekly Wheaties #2314. Maybe there is a future where they work together until either can work independently. And as fast as AI is growing and learning, maybe there’s a chance. Until then, I still stand by a post I made in #2425, noting that The Only EV You Consider Should be a Tesla.
In this same area, Kevin Chen shares Why autonomous trucking is harder than autonomous rideshare. Technically speaking, autonomous rideshare is really easy and very safe. They have a very specific route, or at least an area they must stay within, they typically are in slow-speed areas, and these areas are also friendly to these vehicles with striping, signage, and areas for parking or safe pick-up and drop-off. Similarly, John Deere Reveals New Autonomous Machines to be used in the same way. Fields of crops typically don’t move or change much, so farmers can even map out paths and have the tractors and other equipment retrace the steps.
Maybe at some point, LIDAR systems can help map the world so much that the camera systems can tie into them and help enable full self driving for all vehicles.
🗞️ ICYMI: Tech Headlines
Amazon Is Clearing Out Fire TV Sticks, 50% Off Nearly All Models. I still stand by the Apple TV is the best media streaming device - especially if you are already in the Apple ecosystem. Check out this post in #2503 for more info on cord cutting.
I don’t like to talk about leaks or future coming items, but the new Pebble Watch has been a long time coming. And if you want one, you better Get your new PebbleOS watch now!
US Appeals Court Rejects copyrights for AI-generated art lacking 'human' creator. This isn't the first time, and I’m sure it won’t be the last, but for now - if you want to copyright something, you better create it yourself!
Google’s budget-friendly Pixel 9a is finally here, and it comes with a bold new look. A great buy at a great price. I always suggest the Google line of products if you’re in the Android world, unless you know exactly why you may be buying another device.
😎 POTW: Interesting Maps
I’m sure we’ve all used Google Maps, Apple Maps, Waze, or the like. However, there are really cool things people have found on these maps, or unique iterations of these maps. Without overloading you with too many, here are some ones I wanted to share:
📦 Featured Product
With the weather changing and allergies following trend, perhaps you’re looking for a way to help keep your home’s air cleaner. After researching a few brands, I decided to purchase the Blueair 511i Max. It’s the perfect size for a bedroom or two, works over WiFi with an app that can help track air quality, and is very quiet for how much air it moves - and even quieter in night mode. The filter states it will last 9-12 months (which can be monitored in the app) and helps filter allergens, dust, odor, and dander.
