Tesla Autopilot Update Increases Autosteer Speed, Adds Side Collision Warning



Tesla Autopilot (second-generation hardware) received two more over-the-air updates this weekend.

According to Tesla, Autosteer will not function unless you are traveling 18 mph or faster.

According to Tesla, Autosteer will not function unless you are traveling 18 mph or faster.

As told, Tesla is continuing in its never ending endeavor to release incremental over-the-air updates to its Autopilot system. Another new update, released this weekend, improves the technology even further.

About a month back, Tesla cautiously released the Enhanced Autopilot software to 1,000 vehicles as a test run. A few weeks later, it was installed fleet wide. This current update (for all Tesla owners with second-gen Autopilot hardware equipped vehicles) increases the autosteer speed and adds the side collision warning feature.

The Autosteer feature will now allow for speeds up to 50 mph, instead of the previous 45 mph. According to Tesla, Autosteer will not generally function unless you are traveling 18 mph or faster. However, if you are already using cruise control, you can turn it on regardless of speed. Also, if there is a car detected in front of you, it will allow for speeds as low as 5 mph. But, once you hit 50 mph, the system does not function. This is not nearly as fast as the original first-gen system allowed in the past, but everything is a work in progress, and obviously, safety is at the forefront.

Side collision warning, the other added feature, functions between 7 mph and 85 mph. Tesla calls it a Lane Assist feature that specifically helps in changing lanes. The instrument panel will show lines “radiating” from the side of the Model S or X image if a vehicle is detected to the side of the Tesla, that could potentially cause a collision when changing lanes. This technology has been shown to be a top deterrent of many would-be Tesla accidents.

Source: Electrek

Categories: Tesla

Tags: , ,

Leave a Reply

19 Comments on "Tesla Autopilot Update Increases Autosteer Speed, Adds Side Collision Warning"

newest oldest most voted

5 more mph? I can’t tell if Tesla is trolling it’s customers or just hoping to get 5 PR bumps out of bringing the feature up to the previous (and normal) behavior.

Why do you believe your hypothesis is more likely than Tesla’s statement that it’s to ensure a safer rollout?

unlucky — how it the autopilot working in your car?

It’s working as well as I expected when I paid $0 for it instead of $5000 or $7500. Thanks for asking.

I do find it puzzling that Tesla is being so cautious about increasing the speed at which Autosteer functions in this new rollout, whereas with the first rollout it reportedly worked at up to about 85 or 90 MPH with no restrictions. However, Unlucky, suggesting that Tesla is “trolling” its customers certainly is trolling… from you.

You claim not to be a serial Tesla basher, but multiple comments from you like this one rather strongly indicate otherwise.

That wasn’t bashing Tesla.

Let me go further.

It’s now been over 4 months that Tesla hasn’t been able to ship cars with their 2nd most famous feature working correctly.

This is shameful.

So it wont work in a trafic jam slower than 28km/h at the monent?
Good thing there are continuas updates 🙂

Yes, as there would be a car in front of you for it to “lock onto”

It’s E V O L V I N G !!! 😀

Should be pretty functional / awesome by the time Model 3 comes out. 🙂

I don’t understand why Tesla didn’t program the new sensor suite to mimic AP1, first, and then look to morph-in the usefulness of the extra sensors?

I make such better use of my time with AP1, in stop and go traffic, that I find myself having to be extra vigilant when I go back to the other car. Once you have this option, you’re spoiled. So, I can see why AP2 customers might be feeling teased.

Because they’re shadowing the drives in AP2 mode and collecting data and learning.

They are using a completely new chip so there won’t be much they can recycle or share from AP1 (other than the car UI). Since they have to spend the development time anyways, they probably chose to use the 4 cameras in EAP than try to emulate AP1.

Mobileye’s chip embedded most functionality in hardware. Tesla added some high-level code to combine the adaptive cruise control and lane-following functions into a crude “autopilot”.

NVidia’s chip is much more adaptable, but lacks the basic “built-in” functionality of the Mobileye chip. So this time around Tesla must write a bunch lower level code. I’m sure NVidia gave them sample code and has programmers helping, but it’s still a big task. Furthermore, the NVidia system is set up as a neural network, so Tesla has to develop data tools and training protocols. Virtually none of this was needed for Mobileye.

Thanks. I should have read your comment before writing my comment below. Your comment is clearly much better informed. Speaking as a programmer, it looks like your explanation is sufficient to explain the slow ramp up in speed.

I don’t, either. Too bad Tesla isn’t more transparent about what’s going on. As it is, we can only speculate.

I speculate that the added functions, such as identifying traffic signs, plus the ability to “see” past the car in front to another vehicle beyond, take more processing, and thus slow the system’s reaction time. Since higher speeds impose stricter limits on the time available for the system to react, those two factors present a limit to the system. Perhaps Tesla is just being cautious about decreasing reaction time for its semi-autonomous vehicles, so is doing that incrementally? Perhaps Tesla programmers are working hard at streamlining the processes to reduce reaction time.

Again, that’s just speculation on my part.

Quite frankly, if there is anything tesla doesnt lack it is transparency. For being a private company in the auto industry they are veyr open and transparent. But sure, even more is better thats something I can agree on.

I’m not quite sure what you’re saying but the problem is they overestimated their ability to do what MobileEye did for them in HW1. They are way behind on replicating the functionality MobilEye was providing them.

Computer programming is pretty complex, and different hardware has different capabilities, but I wondered this as well. Paying a ton of money for this feature and a real step backwards to lose that functionality, even if it is Beta.
Feel for all these Tesla drivers who have lost this functionality.

Hopefully, Tesla hasn’t painted themselves into another hardware corner. How do they know for sure that HW2 will be able to get all the way to L5 autonomy?