Photo: @DirtyTesLa via X
Tesla Full Self-Driving has made important advancements in the past months, as the FSD V12.3 has been installed on most compatible vehicles in North America. As many people tested the software for the first time, some claimed that it reacts to hand gestures from pedestrians. Several Tesla owners started testing pedestrian hand gestures to prove (or dismiss) this incredible theory.
Many Tesla owners who tested the latest FSD iteration are convinced that vision-based vehicle autonomy is not only possible but just around the corner. The software has become genuinely good at navigating city streets, and many report driving for hours on FSD without having to intervene. Of course, as advanced as it appears to be, Tesla Full Self-Driving is still a work in progress, with many areas of improvement that need engineers' attention.
However, the rate of progress is expected to accelerate much faster now, especially as the number of people using it has suddenly jumped from about 400,000 to over two million. The number of miles driven has increased exponentially, surpassing one billion miles. There is so much data gold to train the neural networks, and Elon Musk confirmed that Tesla is no longer compute-constrained, meaning progress should be quicker and more obvious from now on.
The enthusiasm made people idealize Tesla's Full Self-Driving, attributing it characteristics that just aren't there yet. To be fair, even Elon Musk was surprised by the FSD progress, which made him overreact. He dropped the Beta moniker and introduced the term "Supervised FSD." At the same time, he instructed Tesla to install the latest FSD software on all vehicles delivered and even those returned from the service. This sometimes meant jumping some required testing steps.
This context explains why some Tesla owners are convinced that FSD has become so good that it now considers the gestures pedestrians make. Sure, the position and direction of movement are important factors that help FSD decide if it needs to stop or go ahead. However, some people were convinced that FSD also responds to hand wave gestures from pedestrians. Specifically, the software would stop when a pedestrian steps in front of the car but would continue if they wave, inviting the vehicle to proceed.
It was intriguing enough that several Tesla owners tried replicating this behavior in real life. It's easy if you have a friend willing to play the pedestrian. The most thorough test was probably done by @DirtyTesla, who tested almost every scenario he could think of. Sometimes, the car seemed to indeed respond to the hand gestures, although, in other cases, it made no difference. It's more likely that the car responds to what the pedestrian does (stops, goes back, etc.) than their hand movements.
Whatever the case, it's clear that understanding other people's gestures is an important part of driving. Autonomous driving systems should absolutely master this type of communication. Often, a small gesture or even looking in the eye of the other driver is enough to proceed in a tricky situation. Also, encountering a police officer while they direct traffic in an intersection should not be an issue for autonomous driving systems.
However, the rate of progress is expected to accelerate much faster now, especially as the number of people using it has suddenly jumped from about 400,000 to over two million. The number of miles driven has increased exponentially, surpassing one billion miles. There is so much data gold to train the neural networks, and Elon Musk confirmed that Tesla is no longer compute-constrained, meaning progress should be quicker and more obvious from now on.
The enthusiasm made people idealize Tesla's Full Self-Driving, attributing it characteristics that just aren't there yet. To be fair, even Elon Musk was surprised by the FSD progress, which made him overreact. He dropped the Beta moniker and introduced the term "Supervised FSD." At the same time, he instructed Tesla to install the latest FSD software on all vehicles delivered and even those returned from the service. This sometimes meant jumping some required testing steps.
This context explains why some Tesla owners are convinced that FSD has become so good that it now considers the gestures pedestrians make. Sure, the position and direction of movement are important factors that help FSD decide if it needs to stop or go ahead. However, some people were convinced that FSD also responds to hand wave gestures from pedestrians. Specifically, the software would stop when a pedestrian steps in front of the car but would continue if they wave, inviting the vehicle to proceed.
It was intriguing enough that several Tesla owners tried replicating this behavior in real life. It's easy if you have a friend willing to play the pedestrian. The most thorough test was probably done by @DirtyTesla, who tested almost every scenario he could think of. Sometimes, the car seemed to indeed respond to the hand gestures, although, in other cases, it made no difference. It's more likely that the car responds to what the pedestrian does (stops, goes back, etc.) than their hand movements.
Whatever the case, it's clear that understanding other people's gestures is an important part of driving. Autonomous driving systems should absolutely master this type of communication. Often, a small gesture or even looking in the eye of the other driver is enough to proceed in a tricky situation. Also, encountering a police officer while they direct traffic in an intersection should not be an issue for autonomous driving systems.
Pedestrian hand signal testing with @WifeDirtyTesla pic.twitter.com/28mOyd7cBG
— Dirty Tesla (@DirtyTesLa) April 7, 2024
Google News