Hey Elon, can these cars self drive in the snow?

April 16th, 2019

---

If not, can I be part of the team that solves that?

And by "team that solves that", I mean the blind camera, 5G from Starlink, In depth GPS paths within 1 mile. Radar as much as possible and assume caked up proxy sensors to at least decide when to stop. Also traction control stopping.

It wouldn't be the primary system. It would be a fail over layer between when cameras are obstructed and when driver is forced to take over.

I would need access to:

• GPS/time Data Collected from Vehicles
• A Starlink satellite providing 5G to Kingston ON +100Km radius (no biggie)
• Self Driving API

I can also aid with dirt road FSD tests.

I can test the driving UX in my current Model 3 and/or the Tesla Pickup (when available).

April 23 2019 Update

So I wrote this before the FSD reveal yesterday. Rather than change it, I think it is best to leave my thoughts out there and focus on showcasing how I was wrong and what I learned.

Reliance on maps is not better than cameras. But I am talking about when cameras are completely obstructed. Or maybe just dashcam enabled.

I guess if dashcam is enabled then sight would still be primary input. So GPS averages are not maybe the best.

I still have a concern with the feasibility of FSD in the winter. But the more I think about it, sometimes I don't drive in the winter because it is shitty. Why should the Tesla feel any different? If the roads have not been cleared after a large snow fall, the Tesla will just get stuck anyway. I believe this amount of snow to be a dead stop with 30cm based on experience.

So, wait for the plows to go by is generally good rule with or without FSD.

As for the cameras being covered in snow. Well, it is safer for them not to, but if the car is assisting you, then as long as FSD doesn't drive you head on into some other car, I think it's a win!

After seeing the FSD investor thing, I admit, Tesla does not need me to help with some inferior GPS average system. That would waste everybody's time.

What I can help with is the edge case crossing at Kingston Mills bridge. It goes down to 1 lane for the bridge. I could help create the stop, look for somebody way the hell down there. If I see cars parked, I might pause a second to make sure they aren't opposite traffic. Then go ahead slowly making sure they are decelerating or stopped or non-existant.

Coming back it would be the opposite, but same rules apply. Wait for oncoming to stop. Apply wait and go algorithm.

Now there is some debate at this crossing. If there are 2 cars, do they either both go, or alternate one at a time.

My opinion (and I have seen frustrated looks from) is that 2 cars should go if they are right behind each other. In face a bunch should go. Alternating is the absolute slowest way to move traffic through there.

Posted In:

jsturgis

ABOUT THE AUTHOR:
Software Developer always striving to be better. Learn from others' mistakes, learn by doing, fail fast, maximize productivity, and really think hard about good defaults. Computer developers have the power to add an entire infinite dimension with a single Int (or maybe BigInt). The least we can do with that power is be creative.

---