As a driver in places where it snows... I'm curious about how it will handle:

1. Fresh unplowed snow where there are no lane markers visible.

2. Significant banks that take up half a lane with two lanes in each direction.

3. Ice (often one doesn't start going through an intersection on a green until you're 100% sure that every car has cleared - I've been on both sides of this, where you don't go because there's a car sliding through, and the car sliding through because slick ice doesn't care about brakes)

4. Hard braking (the "drive slow and use the brakes lightly for a long while")

5. Cold car things. Electric battery life in the cold (heater for the passengers). Wipers frozen to the windshield. Accumulation of snow on the window. Excess snow on top of the car (considered an insecure load).

---

Tangent for the "how do you handle lanes when you can't see any" ... old school approach: https://ieeexplore.ieee.org/document/833362

> Driving a snowplow during whiteout condition is one such example. In order to improve the safety and efficiency of the snow removal operation, a supplemental steering guidance display was proposed and instrumented in a Caltrans snowplow. The guidance information was calculated based on the magnetic markers embedded under the roadway. Operations were successfully conducted on a 6.25 km freeway over Donner Summit on Interstate 80 in California during the 1998-1999 winter season.

https://ahmct.ucdavis.edu/advanced-snow-plow

> 1. Fresh unplowed snow where there are no lane markers visible

Forget unplowed snow. Often in Calgary where I live there's no lane markers visible under the snow for months at a time, even if it's plowed regularly

When all an account posts is claims of first-hand experience with various unrelated technologies, the strategy becomes a bit too obvious: https://news.ycombinator.com/threads?id=hncbw02z5a&next=5009...

Thanks for pointing it out. The phrasing made me slightly suspicious, but those kinds of posts are often hard to detect in isolation.

True. But don't most people drive in the tracks of the car ahead? I know when it snows people sometimes make a three-lane highway into a two-way "rut railway" road, and make do.

I worked on perception systems five years ago and we had de-noising algorithms that are much better now. We would never have dared to drive in obscured lanes, but I assume Waymo’s high resolution maps and sensor fusion are extremely good, but yeah, are the vehicles going to plow through the unofficial lanes?

They've been doing test runs here in Minneapolis for more than a year. You're right that it's a challenge, but I bet they'll do fine after a couple years of refinement.

Yep, full launching in Denver just recently as well.

I have zero idea about lidar operational capabilities, but can you heat them to stop snow and ice from piling up?

Parent is referring to snow in the environment, not on the lidar itself.

When snowing in Toronto you usually cannot see the painted lines on the road, there's just the lines from cars driving through the snow that you follow. Sometimes two lane roads become an unofficial one lane road depending on what feels safe.

The falling snow also bounces the laser pulses(?) back at the lidar. Snow fall can be pretty heavy and make it difficult to see objects far ahead. Not sure how well their handling of rain ports over to snow, but it will be interesting to see!

Damn, you guys have it hard over there

You get used to it. I tried FSD (latest version) last winter in those conditions and it had no issues. Waymo shouldn't have any issues with it either.

Street parking in the core makes what would be a manageable amount of snow into a disaster like 2 weeks a year minimum because plows can't get down older streets.

Sounds like someone needs to develop an autonomous snowblower for narrow streets.

The problem is not plowing the streets, it is moving the snow away from the streets. After heavy snowfalls, plows create huge piles that block a significant part of the road. This is a difficult problem that simply takes a lot of resources to address.

California did back in the '90s.

https://path.berkeley.edu/automation/development-advanced-ro...

It had hands free operation (still needed someone in the seat to control the plow) and lane assist in whiteout conditions when the outside edge of the lane had a cliff (much worse than driving into the ditch).

https://path.berkeley.edu/automation/development-advanced-ro...

> This project is one of the first real-world applications derived from PATH/Caltrans research in the area of automated vehicle control. A modified snowblower with add-on sensors, actuator, computer and driver interfaces was developed, and initial field tests were conducted along the 8 guardrail sections between Kingvale and Soda Springs on the shoulders of Interstate-80 near Donner Summit.

https://path.berkeley.edu/wp-content/uploads/2024/11/develop...

> Various lateral sensing and referencing technologies were investigated for this application; what was found was that machine vision does not penetrate snow, and that the GPS system does not provide sufficient reliability under possible multipath and blockage scenarios. A magnetic-marker-based sensing system was chosen for the initial implementation primarily because of its high reliability and accuracy (better than 1 cm) under all weather conditions. The mountainous highway I-80 near Donner Summit (close to Lake Tahoe) was chosen to be the first field test site. In 2001, magnets were installed along the eastbound and westbound guardrails of I-80, at 4 feet apart and 4 feet away from the guardrail. Binary coding of the magnetic markers was designed (north pole up vs. south pole up) to provide information about guardrail characteristics, such as the shoulder side (right or left of the blower) and the end of guardrail. Eight sections of the guardrail were equipped with magnets for the initial feasibility operations with a total length of 1.46 km (0.9 mile) between Soda Springs and Kingvale.

You just drive over it until it becomes a thick, flat sheet of ice... and hopefully not rutted, or else you'll get stuck trying to move your car in the morning.

Afaik the already do this. But another problem is that falling snowflakes reflect lidar pulses and heating doesn't help in that case

that's in the FAQ. They say they have sensor cleaning tech. But they aren't launching until 2028 so 2 winters of testing should tell.

They will also need mother nature to co-operate. Last winter was brutal in terms of snow, we barely had any snow the year before that!

They linked to https://waymo.com/blog/2025/10/creating-an-all-weather-drive... discussing the snow mitigations all the way from software down to the built-in heaters on the sensors.

That's one of the reasons lidar will prove to be a dead end for this application. It's a temporary enabling technology on the path to the simpler multiple camera based solution.

Best approach would be to use both lidar and cameras. They compliment each other

No, in the limit where everyone has lidar they end up stepping on each other too. Being a totally passive sensor system is a major benefit for mass deployment, on top of the cost advantages.

Lidar is cool when you want to get a point cloud from a fixed location, or dealing with perfectly uniform surfaces (which cars don't because humans can't do this either), but beyond that there are better ways to do things.

“Good morning, LIDAR! You’re looking positively radiant today!”

“Why, thank you, Camera! You are doing a wonderful job of seeing everything in your FOV!”

“Ahh, you are very flattering, LIDAR! It’s an endearing quality in you, for sure!”

LIDAR: blushes

Camera's can't see in the dark. There is no perfect sensor. Waymo relies multiple sensor modalities: Cameras, Lidar, ultrasonics, radar, gps, inertial measurement units.

Cars don't drive in the dark. Roads are usually lit. And every car has its own lights for seeing where you are driving.

Interstate highways and local roads are generally unlit outside of urban areas. But in fairness, Waymos don't operate in those places.

> Roads are usually lit

Not always. Definitely not all freeways.

> And every car has its own lights for seeing where you are driving

Not always far enough. Especially at freeway speeds on unlit freeways.

So how can humans do it?

If humans don't have lidar but can drive it's an existence proof of the viability of that approach.

In addition to vision, humans have proprioception and kinesthesia, hearing, inner ear receptors for balance, a sense of touch, sense of smell. We use all of these sensory inputs when we drive. And even then humans get into lots of accidents.

Lidar is superhuman vision.

How do you use your sense of smell to help you drive? I can't smell anything outside my car thanks to the air filter.

The real answer is very simple: safe drivers slow down to adapt to poor conditions (including visibility). Lidar might reduce how much it needs to slow down but I don't think it will be eliminated because you can't drive with only lidar.

> How do you use your sense of smell to help you drive?

I smell smoke before I see it. Useful if there's a forest fire up ahead.

Burning rubber or smoke from the engine is an indicator of a mechanical issue.

> you can't drive with only lidar.

Nobody said you could. This is a strawman.

Tesla stans are the only ones trying to convince everyone you can only drive with cameras. Everyone else is pretty united on the usefulness of multiple inputs.

> Nobody said you could. This is a strawman.

Didn't mean it in that way. I meant a car with camera+lidar would realistically still slow down because of the reduced camera visibility.

> Tesla stans are the only ones trying to convince everyone you can only drive with cameras. Everyone else is pretty united on the usefulness of multiple inputs.

Ignore them. The latest Model S and X have radar again, but possibly unused right now, so they are probably going to roll that out in a refresh of the 3 and Y eventually.

Humans can't do it safely during a proper blizzard. (Toronto doesn't really get proper blizzards, though).

Also microphones. Don’t forget the (external) microphone sensors.