Support CleanTechnica’s work through a Substack subscription or on Stripe.
I will be turning 85 this month, so I definitely don’t mind getting some help with the driving. It’s especially appreciated when I’m tired on the long drive home at night after a concert in Salt Lake City.The navigation appears to remember my driving history and present the correct destination at the top of the list. For example, it knows we go to church first thing Sunday morning, so it places the address of our church at the top of the list of my navigation choices. Even better! In my Tesla Model 3 with Full Self Driving Supervised V12.6.3, I can verbally tell the navigation: Navigate to the LDS Church, Navigate to the Jailhouse Saloon, or Navigate Home. More often than not, it understands my command even for obscure locations.
Then I’m still blown away how Tesla’s FSD Supervised very accurately drives my car to that destination perfectly without intervention more often than not. It stays in the lane more accurately than can I do at my most attentive. It makes nice wide left-hand turns at intersections so that it can avoid traffic from the other direction. It fully stops at stop signs and waits exactly one second before proceeding if there is no traffic. I would probably do a rolling stop in low traffic areas, but what it does is strictly legal and takes little extra time. It waits for cross traffic at intersections and proceeds briskly immediately after the last passing car. It moves over for parked cars, pedestrians, and bicycle riders and will drive around a puddle in the road. It has numerous times braked for deer crossing the road. Recently, it avoided fallen logs that obstructed part of the travel lane. It now responds correctly to traffic cones that close off a lane of the highway. On a super highway, it observes the slow-moving truck well ahead, moves into the passing lane, and quickly returns to the right-hand lane before an impatient fast-moving driver behind me is tempted to pass on the right. Although, lately, it doesn’t return consistently.
Some FSD Supervised faults
FSD Supervised observes stop signs, speed limit signs, and traffic lights, including turn signals, and responds appropriately. You get additional reassurance that it understands because it not only sees the signs and lights but also presents the correct icon representing the sign or light on the screen. However, my Tesla does not yet slow down for 20 mph school zone speed limit signs with the big yellow alternating flashing lights. This is baffling to me! My only imagined explanation is that Tesla just hasn’t gotten around to programing the system to do it. But they have a team of hundreds of software engineers working on FSD and they haven’t had time to correct a glaring fault like this? Maybe there are no school zone speed limits in California or Austin, Texas.
When it comes to correcting other faults, it’s hard to imagine solutions. Remember my car neatly dodged a puddle in the road and braked for a deer. However, it also braked for some big tar stripes in the road, which was not appropriate. We call this phantom braking. It even happens quite frequently when we don’t observe anything that could cause it. How to get the system to differentiate between true dangers and phantoms? I have no idea.
Perhaps the most serious problem with FSD Supervised is the occasional failure to get in the proper lane prescribed by the navigation at a stoplight or stop sign. Sometimes it will then even make a turn not prescribed by the navigation. However, the good news is that it will recover, just taking a longer route to your destination. Why? Perhaps FSD needs to have every intersection in the US in its database with the function of each lane. The navigation system already has every road in the US, and most likely every curve with the curve-safe speed limit, in the database. It wouldn’t be that much additional work to include each intersection and the function of each lane. As drivers, we sometimes get in the wrong lane because we can’t see the pavement markings ahead of a long line of cars.
Early versions of FSD wouldn’t make very sharp turns like those marked by a 15-mph speed limit and the turns required in a rotary/roundabout. Recent versions handle those situations perfectly. However, there is a turn in my neighborhood, Bonkowski Road onto Voyager Road, that is only ~15 degrees coming into the neighborhood. That makes the turn going the other way ~175 degrees. FSD wouldn’t make the turn. Also, FSD frequently gets confused exiting the huge parking lot of a certain big box store. Additionally, you need to be in Drive to initiate FSD — thus, it won’t back you out of your garage.
Bottom Line: There are still numerous situations that prevent you from getting into your car in the garage, entering the destination, and having your car drive you there while you read a book or take a nap.
Will Tesla maintain the software even on its older cars? I’m driving a nearly 6-year-old 2019 Tesla Model 3 which has L3 cameras and computers, but Tesla has recently upgraded their new cars with more capable cameras and better L4 computers. I have had 6-year-old Apple computers where Apple stopped upgrading the operating system. I found that I could no longer back up my new iPhone to my computer. In the other direction, I couldn’t update the operating system if I still wanted to run my favorite old version of the Final Cut Pro video movie editing software.
This brings up the question: how long will Tesla maintain the software in its cars, especially when its new cars have more advanced computer hardware and cameras that have better prospects for full automation? Also, will Tesla ever be able to honor its promise that all L3 hardware cars will eventually be able to perform at L6 (full) automation. It will be very tempting for Tesla to concentrate its automation efforts on cars with the improved cameras and computers. However, Tesla has millions of cars out there like mine with L3 hardware, and you can’t expect people to abandon their cars as quickly as they abandon their computers.
I am reassured that Tesla is still offering software updates every few weeks. I am also reassured when I see actual improvements in performance with the updates. One small example: With the previous version of FSD, or even traffic aware cruise, my car would follow the car ahead annoyingly close. It seemed like not much more than 10 feet. I was not comfortable with this and had to reduce my speed manually. I am very pleased to see that Tesla has responded to our numerous complaints and increased the following distance to at least 20 feet.
Do I have the latest version of Tesla’s FSD software?
No! Back in 2019–2020, Elon promised that anyone with the latest cameras and HW3 computer would be able to rent their cars out as robotaxis within 12 months. At that time, Tesla agreed to upgrade owners with HW2 computers to HW3 computers at no cost. 5 years later, we are no closer to full Level 5 automation on any Tesla cars than we were 5 years ago. Not only that, as mentioned above, Tesla has recently been shipping cars with improved cameras and more powerful HW4 computers. However, owners of HW3 computers like myself are stuck on FSD V12 with no upgrade path for the better computers and cameras. HW4 computer owners are now running FSD V13. However, it is reassuring that we are still getting software upgrades with improvements in FSD performance. It is also reassuring with schadenfreude to hear reports that FSD V13 users are not getting any quantitative improvement in performance.
My Tesla Model 3. My wife and donkeys. Custer State Park, South Dakota. March 23, 2023. Fritz Hasler Photo
A few items for those reading about these things for the first time
Aggressiveness setting plusses and minuses
The aggressiveness toggle has three settings: Chill, Average, and Hurry. I’m amazed how aggressively FSD finds a small spot between cars and changes lanes on a busy freeway. This is needed when you need to move right in time to exit on a multilane highway. However, I am very annoyed that I can no longer get the over-speed amount that I set. I set the over-speed setting to 10 mph, but even in the Hurry mode, it never reaches 10 mph over like it did in earlier versions of FSD.
How does Tesla know that drivers are paying attention to the road?
Since the earliest versions of lane assistance software, Tesla has required drivers to apply a small amount of torque to the steering wheel to let the system know that the driver was paying attention to the road. This is a flawed system because the driver could be reading a book or texting with one hand and still apply the torque. However, Tesla has a much better way to monitor driver attention. There is a camera just above the mirror that observes the driver’s face and eyes. Tesla is very strict with this system. If you look away from the road, look at the navigation screen or look down at your phone for more than a few seconds, you get a blue flashing warning at the top of the screen. If you persist with prohibited behavior, it is followed by a full-screen, flashing-red steering wheel and finally disengagement from FSD. The system works great in daylight as long as the driver is not wearing dark sunglasses. I have found that I need to keep a hand resting on the cross bar of the steering wheel at night when the camera can’t see my face.
Why I still appreciate driving my ~6-year-old 2019 Tesla Model 3 EV with 150,000+ miles
I am very spoiled by the instant, whisper quiet, very strong acceleration I get in my dual-motor Model 3. And I don’t even have the Performance model. When I am pulling my 2500 lb ski boat out of the water at a boat ramp, I can barely feel it behind me.
Referral Program: If you are buying a new Tesla, Tesla has reactivated its referral program. If you find any of my articles helpful to you, please use my referral link: https://ts.la/arthur73734 (be sure to use it when you make your order). If you are buying a new Tesla and use my link, I believe you’ll receive $1,000 off the purchase price of a Model S or X, or $500 off the price of a Model 3 or Y. You will also get 3 months of Full Self-Driving (Supervised). Just be prepared to intervene immediately if it screws up.
Sign up for CleanTechnica’s Weekly Substack for Zach and Scott’s in-depth analyses and high level summaries, sign up for our daily newsletter, and follow us on Google News!
Advertisement
[embedded content] [embedded content]
Have a tip for CleanTechnica? Want to advertise? Want to suggest a guest for our CleanTech Talk podcast? Contact us here.
Sign up for our daily newsletter for 15 new cleantech stories a day. Or sign up for our weekly one on top stories of the week if daily is too frequent.
CleanTechnica uses affiliate links. See our policy here.
CleanTechnica’s Comment Policy
Share this story!