Live data from Hacker News

Redefining what a map can be with new information and AI

blog.google

201–210 of 293 posts

Re: Redefining what a map can be with new information and AI

#201
post #59

Earlier quoted context omitted.

The quoted paragraph clearly states that the lower-emissions route will only be given by default when there are multiple routes with a similar ETA. So it will never tell you to take a route that takes longer just to reduce your emissions, without also telling you about the shorter route with higher emissions. Since you have seemingly already abandoned google maps due to their activism before reading this story, what…

How is the fastest route not the least carbon footprint?

The only thing that matters for your car's carbon footprint is how much gas you use. So regardless of time, if you have more gas in the tank at the end of your trip, your carbon footprint is lower.

The faster you go, the more your fuel efficiency decreases. For the most part this is due to wind resistance, because there is a velocity^2 argument there(ie if you double your velocity, you quadruple your wind resistance). Your engine is also designed to operate more efficiently at certain speeds but that is less directly quantifiable than wind resistance.

So you might be able to get 40MPG while doing a steady 40 mph, but only get 20MPG when doing 80mph.

So if you need to go 40 miles, if you can do it at 80mph it will take you 30 minutes, and will cost you 2 gallons of fuel. If you do it at 40mph it will take you 1 hour and cost you 1 gallon of fuel. So in this contrived example the longer trip has a lower carbon footprint.

In the real world, there are rarely two parallel roads going exactly the same place but just with a different speed limit. Usually you need to drive out of your way to get onto a highway to make your trip go faster. On the other hand driving directly at a slower speed usually entails more starts/stops at traffic lights which eats into your fuel efficiency. Google is probably in a good place to answer the question of how fuel efficient is a given route because they have so much data about average speeds, accelerations/decelerations required/ etc.

Re: Redefining what a map can be with new information and AI

#202
post #188

Earlier quoted context omitted.

Driving a Chevy Volt (with an 18 kWh battery) makes you very aware of the different fuel-costs of different routes -- I very often have a choice of driving "over" or "around" small mountains (~hundreds of feet of elevation change, not ~thousands), and you can easily see the battery drop as you go up, up, up and it doesn't always recover much going down, down, down.

Going up and down mountains should be mostly a wash, apart from these small losses. What's really eating into your battery in a way you can't recover with regenerative breaking ever is wind resistance. Routes that make you drive faster for longer cost more energy. (And given the way wind resistance works, for the same average speed, a variable speed is more expensive than a constant one.)

Optimal regenerative braking recovers like 2/3 of the energy lost, and suboptimal braking -- for instance, if you need to decelerate more rapidly because of a bend in the road -- is worse.

Every 100 feet of elevation change amounts to ~1% of the battery capacity of a Chevy Volt; a couple of small mountains, with lots of ups and downs, and it's easy to lose 5-10% of your range versus taking a more level route, 5% if you brake well, 10% if you brake poorly.

[Which is not to say that hitting the thruway doesn't also kill your range fast.]

Re: Redefining what a map can be with new information and AI

#203
post #76

I find it more believable that Google attracts engineers who want to save the environment than Google is secretly manipulating the routes of the masses for profit.

> than Google is secretly manipulating the routes of the masses for profit That's the first I hear of this conspiracy, are there really people that believe this? FUD has honestly gotten out of control into very strange territory.

Take a peek at some of the other comments on this thread hinting or outright suggesting it.

Re: Redefining what a map can be with new information and AI

#204
This is the opposite of a map. A map helps you build a mental model for how you navigate space by showing you a broad view of the land, potentially enabling you to navigate to anywhere. This offloads the effort of learning, such that you don’t learn to navigate the space yourself, and only gets you from one point to another.

Re: Redefining what a map can be with new information and AI

#205
post #3

So it'll only show walking directions?

I get that you're being flippant, but I did have a coworker who calculated the carbon footprint of his commute for various modes of transport, including walking. He came to the conclusion that, given the average amount of CO2 produced per calorie consumed, riding an e-bike was actually the lowest-impact way.

That assumes you would be eating more to compensate for the activity. Many people, because they spend so much time in their car or just being sedentary otherwise, actually need to go out of their way to exercise (e.g. in a treadmill in a gym). And for people who don’t already exercise, the marginal increase in CO2 is probably worth it for society.

Practical example: I primarily walk and cycle, used to have a car but no longer do. I need to get 1-1.5 hours of walking in a day for my health. I chose walking because it is specifically calorie inefficient and unlikely to make me hungry like running or cycling.

If I can walk for all my business, I no longer need to take a separate walk in the evening to get my exercise. If I drive a car or ride the e-bike (the e part robs you off fitness benefit) I now need to take a separate dedicated walk.

If you consider that the CDC wants every American exercising several hours a week, I think your friend’s analysis would come out differently.

Also the analysis should factor in the reality that walking 6 miles will likely not trigger the same level of calorie intake increase as running or biking 6 miles; it is very easy to e.g. run and actually consume far more than you burned due to appetite response.

Re: Redefining what a map can be with new information and AI

#206

Earlier quoted context omitted.

At least for cars, it makes little sense because the dominant factor is going to be the time on the road, which they in theory have been trying to minimize already. Any additional micro-optimizations are going to be fractional.

Unless of course they optimize for me not driving under a train bridge with my camper. That would be a big help. I wish they took into account the vehicle you are driving.

Can you elaborate? I'm not sure I understood. I've never driven a camper.

Re: Redefining what a map can be with new information and AI

#207

This is the opposite of a map. A map helps you build a mental model for how you navigate space by showing you a broad view of the land, potentially enabling you to navigate to anywhere. This offloads the effort of learning, such that you don’t learn to navigate the space yourself, and only gets you from one point to another.

You are right, but most people want navigational aid, not map. Maps were popular because they were good navigational aid, not because they help building mental model. If better navigational aid becomes available, most people will gladly abandon map's mental model building feature.

Re: Redefining what a map can be with new information and AI

#209
Sometimes I feel like the advertising dollars I give Google get put into a nameless machine. Sometimes I feel like they come out, converted to magic that the world benefits from, such as Google maps. It's been one of the biggest use cases of tech for me in my life. In awe that never again I might stand frustrated in the middle of an airport looking for my gate when I've only a few minutes left to get there.

Re: Redefining what a map can be with new information and AI

#210
post #167

FTA but not the main point of this thread: we’re now able to bring Live View to some of the trickiest-to-navigate places indoors: airports, transit stations and malls. Seems to me that these are the easiest places to navigate indoors, because of the plentiful signage. Malls probably the hardest of the three, but still pretty easy IMO. I've never had any issue finding a gate or baggage claim at an airport, since there…

A lot of bigger underground transit stations are literal mazes! Even in the somewhat smaller ones in e.g. Tokyo, you can easily get lost as a tourist or first-time visitor. I'm sure this will be valuable to people especially with accessibility needs.

I definitely could have used some indoor navigation when trying to find stuff underground in Japan, like looking for T's TanTan in Tokyo Station (amazing ramen, try it if you get the chance).

Took me a while to figure out it's actually behind the ticket booth, and that you can only access it if you're taking a train (or have bought a ticket just for ramen).

Post reply on HN