Tesla Full Self-Driving on 45 Minute Commute!

Tesla Autopilot Full Self-Driving on my 45-minute commute in 2020

Here's how my Tesla Model 3 (HW3 + Enhanced Autopilot & Full Self-Driving) drives in the dark and rain on my daily commute in 2020. We also take a look at the latest Tesla software update that added automatic stopping at stop lights and stop signs to see how it performs in real life.

🚘 TESLA RESOURCES

⭐️ FAVORITE RESOURCES

👇 CONNECT WITH ME

Tesla develops and deploys autonomy at scale, and they believe that an approach based on advanced AI for vision and planning, supported by efficient use of inference hardware is the only way to achieve a general solution to full self-driving.

Hardware
Build silicon chips that power our full self-driving software from the ground up, taking every small architectural and micro-architectural improvement into account while pushing hard to squeeze maximum silicon performance-per-watt. Perform floor-planning, timing and power analyses on the design. Write robust, randomized tests and scoreboards to verify functionality and performance. Implement compilers and drivers to program and communicate with the chip, with a strong focus on performance optimization and power savings. Finally, validate the silicon chip and bring it to mass production.

Neural Networks
Apply cutting-edge research to train deep neural networks on problems ranging from perception to control. Our per-camera networks analyze raw images to perform semantic segmentation, object detection and monocular depth estimation. Our birds-eye-view networks take video from all cameras to output the road layout, static infrastructure and 3D objects directly in the top-down view. Our networks learn from the most complicated and diverse scenarios in the world, iteratively sourced from our fleet of nearly 1M vehicles in real time. A full build of Autopilot neural networks involves 48 networks that take 70,000 GPU hours to train. Together, they output 1,000 distinct tensors (predictions) at each time step.

Autonomy Algorithms
Develop the core algorithms that drive the car by creating a high-fidelity representation of the world and planning trajectories in that space. In order to train the neural networks to predict such representations, algorithmically create accurate and large-scale ground truth data by combining information from the car's sensors across space and time. Use state-of-the-art techniques to build a robust planning and decision-making system that operates in complicated real-world situations under uncertainty. Evaluate your algorithms at the scale of the entire Tesla fleet.

Code Foundations
Throughput, latency, correctness and determinism are the main metrics we optimize our code for. Build the Autopilot software foundations up from the lowest levels of the stack, tightly integrating with our custom hardware. Implement super-reliable boot loaders with support for over-the-air updates and bring up customized Linux kernels. Write fast, memory-efficient low-level code to capture high-frequency, high-volume data from our sensors, and to share it with multiple consumer processes— without impacting central memory access latency or starving critical functional code from CPU cycles. Squeeze and pipeline compute across a variety of hardware processing units, distributed across multiple system-on-chips.

Evaluation Infrastructure
Build open- and closed-loop, hardware-in-the-loop evaluation tools and infrastructure at scale, to accelerate the pace of innovation, track performance improvements and prevent regressions. Leverage anonymized characteristic clips from our fleet and integrate them into large suites of test cases. Write code simulating our real-world environment, producing highly realistic graphics and other sensor data that feed our Autopilot software for live debugging or automated testing.

As an Amazon Associate & partner for other affiliate programs I earn from qualifying purchases

[TKCIW]

1:40 it’s funny you say this. I’m a first time Tesla owner (Model Y for about 3 weeks now) and I utilized Autopilot for the first time in heavy rain at night. My GF was sitting passenger and we BOTH agreed that autopilot is this condition was significantly better than what we could comfortably do ourselves. I could barely see lanes thanks to the misty rain closer to the ground, yet autopilot worked flawlessly and left us impressed! I agree with you 100%!

[Sibs]

I've always been super impressed with how well autopilot handles rain & weather. I've been driving with really low visibility in rain/fog and autopilot sometimes sees cars before I do. That radar & camera combo is incredible!

[arvindnarayanan]

10:22 I kinda agree with Tesla's logic when the lane merged and I also while driving try to drive a little bit to the left. This is essentially to ensure that if at all there is a vehicle behind you (on the faster lane in this case) which might be faster and was planning to overtake you and then merge, us staying slightly to the left just avoids such tricky (and potentially dangerous) situations. It then forces the car behind to stay behind and hints to them on not to overtake. Just one of those unspoken mind games to ensure people play safe.

[martin]

You should use better wiper blades or a faster wiping intervall, looks like you can see nothing through the windscreen

[bobbyjayy_]

This was actually a really good video! Watched it thru and thru! Will hopefully be getting a model 3 performance + FSD within the year, I’ll probably watch this right before for a refresh! Stay safe

[mingye]

I can definitely attest to using autopilot during bad weather where it’s visually difficult to see the roads clearly. Autopilot is my go-to function to help. Big safety feature!

[Hawkman]

6:28 Answer to why doesn't car move out of fast lane: my observation was that it also depends on whether you have a car behind you, as well as your higher set speed. Noticed it many times on 1k journey back from Seattle last weekend.

[franksmith]

The reason it stayed in the fast lane (around 6:20) is because you were passing and Off ramp, so therefore it knew that there was an On-ramp coming up and it was waiting to see if it was clear. Normal driver management (well here in the UK)

[MrDancemeh]

I also live in Louisville & it’s so cool seeing this car gracefully travel thru the awful roads I know so well

[BrysonWooden]

Excellent video. Great job. I'm looking forward to owning a Model Y hopefully this year if not next.

[SpencerOlson]

This is really cool. I want one so bad!!!

[happyjohn]

You drove 45 minutes in 17 minutes. Honestly, that's pretty impressive. Pulled a Han Solo!

11:35 AM
9/7/2020

[jakecollier]

that one lane road with the oncoming traffic looks terrifying

[alextorchia]

i don’t know why, but watching this was just so calming

[AniBanerji]

One of the best videos on auto pilot, thank you so much and I just subscribed to your channel coz I was super impressed. Great job!

[karl_wiley]

Was curious what the white bar above the blue autopilot designation was. After watching for a second I realized it's the acceleration/deceleration bar. Awesome.

[RobbieGVT]

I always appreciate your level of thoroughness in all your videos. Thank you!

[Goose]

Great video Andy! I have EAP on my M3 long range dual motor but haven’t pulled the trigger on FSD yet. Navigate on Autopilot is a game changer on the interstate, and I use it every time I’m on the interstate.

[ryansheehan]

I’ve been a skeptic of FSD until I started watching these videos. Now I’m much more confident about it.

[Kiddo]

You have a very nice, relaxes narration style. cool video, can't wait till i can afford a tesla.