Hacker News Reader

Yes, but could we maybe get a decent approximation of the windshield temperature from interior temperature, exterior temperature, and speed? Combined with interior/exterior humidity (maybe these gauges might be the expensive/finicky part of this project?), we could calculate a probability of window fogging.

It doesn't have to be perfect -- we can have it just turn on the defrost whenever the probability of window fogging is >10% or something.

You could do one of two things. First, a heated windshield like I believe some Landrovers have. It’s about $5000 to replace, last I heard, but it works. Second, a double pane windshield like house windows or some motorcycle helmet face shields. These work 100% of the time, but would definitely be more expensive as well.

Best car I had for window defrosting/demisting was a Ford Mondeo which had a fine-mesh heating element embedded. Super fast clearing. More expensive than regular glass but not $5000 by a long shot.

I had a Honda Accord with a cracked radiator.... It defrosted right quick, but idling it after it warmed up was very ungood. After the radiator was replaced defrost went back to normal.

You can get it for most Fords. It's one of my favorite features, and any car maker that caters to the more northern people should have that option.

When it's sort of cold, but not "ice on the car" cold, it works great as defogging. Way faster than waiting for the car to heat up. When there is ice on the car, you turn on the heaters and within a minute or two you can simply use your wipers to clear the window of ice.

Could glue/epoxy a sensor right behind the centre mirror I’d think. It’s far enough from the defroster vents that it wouldn’t heat up quickly from the hot air.

But, figuring out if visibility is good through the windshield is still manifestly a much easier problem to solve than self-driving cars.

It's not super accurate, but most cars these days also have outside temperature. Add a humidity sensor inside and outside, and we have temperature and humidity inside and outside the car, which should give us enough information for a basic microcontroller to approximate the temperature differential across the glass and whether or not fog (or frost) is likely.

I think knowing the temperatures and such amounts to predicting the foggy state. Detecting it could look much different. Visual observation is obvious. With purpose built glass, perhaps one could look at electrical properties on the inside of the windshield. Maybe changes in the reflectivity of the glass could be used. Maybe some refractive index shenanigans?

This seems like a simpler route. Why not just a camera pointed at a test mark on the windshield that checks if it is visible through the glass?

Or simply look for reflectivity at an angle, like the recent HN story about the guy who implemented a touchscreen on his Macbook by adding a small mirror to the camera above the screen.

> In order to detect it you'd need to know the glass's surface temperature ideally in the middle (away from the car's body) and also know the interior temperature

Or, perhaps more simply, have an optical sensor that detects and reacts to the fogging itself sooner than a human would.

Wouldn't it be easier to point a camera at the windshield and detect a change in opacity? I would think that's a far easier problem to solve than, say, facial recognition.

But figuring out the glass's temp is non-trivial.

Not really. Many windshields already have elements embedded in them like the ultra-thin wire used for the AM radio antenna. Sensing temperature via the resistance of a wire like that is about as trivial as it gets.

What if the glass had some small heating applied to it constantly?

Complete waste of energy? Additional energy required for AC when it is hot?