The article focuses a lot on street images.
It's worth noting the existence of the open source street image platform https://panoramax.fr.
Disclaimer : I'm building a local open source "clone" of Google Maps, on the Web, for France. https://cartes.app. Code : https://github.com/cartesapp/cartes
I have a front-facing dashcam in my car. It records videos when I drive. Would this be useful for others? The videos are not stamped with GPS coordinates but I don't go off road or anything so you can usually tell the name of the street or highway I am on just by the street signs. Are these dash cam videos worth saving?
Edit: this is in the US
I contribute street images to Mapillary and Google Street View. Mostly in nature /trails where the cars don't reach, with my goal of making it easier for people to plan their journeys and get outside. I've never quite understood how to contribute to Panoramax, though?
No shared instance for uploading images for Norway as I see it, and feels weird having to set up my own federated instance if all I want to do is offload all my pictures for others to use? I know the one on the page you linked says "OSM World wide", but reading at the wiki https://wiki.openstreetmap.org/wiki/Panoramax it says not to use it for things outside france, and I also can't understand how I contribute to MapComplete.
(here's the tooling I've built for myself btw https://github.com/Matsemann/matsemanns-streetview-tools/ )
I believe the OSM instance is usable outside of France. Checkout the map here : https://panoramax.openstreetmap.fr/#background=streets&focus...
Yeah, but given the quote from the OSM wiki I'm a bit apprehensive about spending time uploading a couple of terrabytes / hundreds of thousands of 360 images. Either because it's not worth it (no one here will ever see them, or they may be removed), or because it would be detrimental to the .fr project for me to waste their resources.
Thanks. I've sent a private message to one of the project owners.
cartes.app looks sharp!
Thanks ! Lots of polish and features left for a proper v1, on the scope of France. Then we'll think about translation and i18n.
The Grab engineering blog is pretty high quality, here's a related article from a few years ago that I liked: https://engineering.grab.com/correcting-restaurant-locations...
Nice to read they are significantly contributing back to OpenStreetMap.
I'm currently in Pokara, Nepal and Google maps has definitely hallucinated a lot of streets that don't exist (computer vision false positives I presume) This makes Google useless for routing because it sends you into dead end alleys and driveways it thinks cross through to the next road.
Organic Maps (OSM client) on the other hand is much more accurate and supports offline routing.
Organic Maps or OSM-based maps in general are great, but for routing nothing can replace Google Maps (or Waze) here yet, because they are the only ones with actual, near real-time traffic data.
And here in Belgium, your 100 km trip could take anywhere between 50 minutes and 2 hours depending on whether there's traffic in the Kennedytunnel or an accident in Nazareth. It's crucial for me to check traffic before I leave, so if there's a one-hour traffic jam I can just do whatever else for an hour before leaving.
And I haven't found a way to check that without Google Maps yet. But I use OsmAnd if I'm not taking the highway.
I’m surprised they can’t flag these misclassified roads by tracking traffic flow through them. A bad road should exhibit little traffic (or even more reroutes).
Just simple neglect. Not enough revenue from the region to warrant operation and engineering time.
Both good and bad roads in remote areas will likely appear to have zero traffic flow data as there is no cell phone coverage.
Even in Europe Google Maps can be surprisingly bad. For example, in Portugal, their routing algorithm thought you could drive across the top of this dam. But when you get there, you absolutely cannot: https://maps.app.goo.gl/jBRNbeqZiY9cu8eL9 If you zoom out, you'll see it's quite a long detour to backtrack.
(I reported this mistake, and it seems to be corrected now)
Could it be that motorcycles can drive there? I get routing errors where maps will show a turn where there's only a divider and a pedestrian size opening. I notice motorcycles taking a shortcut there though.
This is common in South Asia. Even cities like Chennai have this happen. I was driving down a road that became narrower and narrower until I had to back up or end up like a character in a Junji Ito manga.
This happened to us in Minorca. It was a pain to backup 50 meters with only 20cm of clearance.
My company, http://www.mosaic51.com, has been doing this for about 6 years now. I have been deeply involved in 360º imaging for over 20 years.
We are making the shovels for the AI gold miners now. You all need data. We can make that visual training data.
Let's build a large spatial model together? Hit me up!
Mildly interested to see if they come up with anything useful regarding cultural map differences.
An ongoing PITA in Thailand was Thais being absolutely baffled by Google Maps. You couldn’t hand your phone with an pin to a driver and expect it to make sense. All the taxi drivers and motorsai guys have smart phones — hell I even saw one at the _win_ using a VR helmet — but were totally perplexed by Google Maps. Weirder still was the insistence at places like Ikea that you drew a map to your property, but the map they wanted needed to be entirely landmark based.
I feel like there was some underlying sociological issue there where Thais thought of maps in terms only really of landmarks, and not as birdseye views of territory. Would be fascinated to know what Grab’s experience of it is, and also if Grab and friends are slowly converting professional drivers to real maps.
To be fair, I find modern Google maps absolutely perplexing as well, the user experience has really degraded as ads and other functionality has been shoved in on top of what I want - a map to get to where I'm going.
Instead it's so damn slow as to be hard to get anything done while on the go, even with maps downloaded, modern Android phone, etc.
It really, really needs a simplification refresh.
I just learned from this youtube video: https://www.youtube.com/watch?v=_HSYTIEXa5w that google localizes it's driving directions to be landmark-based in certain countries. In this video it mentions India specifically.
Why would Grab need to "convert professional drivers to real maps"? According to the article, it's already solving for all the unique problems by building out their own map.
Your reads to me like you expect that if someone comes to you and gives you GPS coordinates, you should be able to know where that is because you have a GPS in your pocket. No, you'd need to put the GPS coordinates into something and have it translated to something that made sense for you, like into Google Maps. Same goes for Thais, even if it's a PITA for you.
Unique problems require unique solutions, and I think your definition of a map is probably too closed-minded. Imagine trying to take a real map and deliver something here: https://www.youtube.com/watch?v=wQyVMy6_n1k
While your point is certainly valid, the parent's experience lines up with mine. There is a disconnect for rideshare/delivery drivers even between getting instructions from Google Maps vs. me reading out instructions straight from Google Maps. If I say exactly what Google Maps, then we're able to get to where I need to go.
I will say that it is changing, though. For younger rideshare drivers, there is a greater reliance on and ability to use Google Maps. In my experience, it is older drivers that have an interesting time adapting. This makes me think that there is certainly some ostensibly unrelated cultural norms at play: for example, Thailand exhibits a high degree of social hierarchy dependent on age, and a priority on saving face. Totally ignoring Google Maps, if I know the best way to get somewhere in Bangkok, and I know that most taxi drivers don't, me communicating it in Thai as respectful as is culturally appropriate can and does get shut down or ignored. In these instances, we invariably get lost and/or stuck in traffic.
Is a dislike for Google Maps in navigation completely a function of cultural values? Probably not, but this kind of technology isn't thought to be by default an amazing convenience in many parts of the world.
Very interesting, thanks for sharing.
> Why would Grab need to "convert professional drivers to real maps"?
Because Grab have been around a long time and their custom maps are pretty new
> I feel like there was some underlying sociological issue there where Thais thought of maps in terms only really of landmarks
This is such an interesting comment, I'm curious of your age. I grew up and learned to drive before smart phones existed and landmark based directions were extremely common. Directions like this were expected, moreso than street name based directions. "Take a left at the 3rd light after the freeway, then a right after 7-11, and it is the 5th house on the left with the blue door and red Truck in the driveway".
What is interesting is that these drivers are still using this older method.
I've seen both delivery drivers (on motorcycles) and Grab taxies use GPS navigation in Thailand. Regular taxi drivers might have an ego thing or something. Or maybe they can't read the latin script.
Grab also has Uber-like service doesn't it?
Grab is basically Uber + Ubereats in south-east Asia
(Quite literally - Uber sold off their SEA operations to Grab in exchange for a stake in Grab, when Uber exited the SEA market)
Grab is also a bank
Yes. And they accept cryptocurrency natively (without using gift cards), extremely practical for believers in the space.
Uber also owns a significant share of Grab.
More crucial to their success was that Grab accepts cash payments. This alone was the reason why Grab succeeded and Uber failed in SEA.
Uber always had cash options in SEA and still does in countries it continues to operate in, like India .
Each country has and always has a lot autonomy on features and services offered . Just their booking methods in different airports can have lot of variation , OTP workflows in geos with high no show rates , transportation options like motorcycles or Tuk tuk or public transit integration all are unique to many locations
Consolidation and exits happened for variety of reasons i strongly doubt it was ever because of product features or lack thereof
Interesting because when I traveled in India they had cash as an option for Uber. I didn’t try UberEats though.
Uber Eats exited India long back.
Cash is not only accepted but in specific Indian cities expected , I.e. the driver will call you and if you don’t offer cash they will cancel the booking or won’t show up at all .
Can you elaborate? In the U.S. you place an order and are quoted a price which you pay (typically with a credit card) before your order is sent for delivery. Receiving a call from the delivery driver about paying in cash would feel like an extra fee. In the situation you describe is no payment made in advance?
As you hypothesize it is about extra fees and renegotiating the price. Common tactics include asking more than the price shown on the app, or refusing to give change back,or pretending not to have any etc.
The commenter was talking about cabs and autorickshaws, not food delivery drivers.
Uber and others pay out on a weekly basis. Cabbies would rather have the cash now. And by cash, they mean a direct payment outside the app. Typically this would mean a direct transfer via UPI.
Its main services include ride-hailing, food delivery, payment and digital bank and more...
I gather it's kind of a "super-app" for its region, similar to WeChat in China? (Or what Elon Musk wants X to be, but so far isn't even in the ballpark...)
Kind of, but not exactly. Super-apps like WeChat have a huge number of independent apps within it, operated by third-parties. Apps like Grab are more like different divisions within the same organisation. So Grab is more like Uber + Uber Eats + InstaCart all operated by the same company, inside a single app. But sometimes people call things like Grab super-apps as well. I think it’s probably wrong to describe something as a super-app unless it has third-party apps within it.
X has nothing to do in this comparison.
Use Grab a lot and I don't consider it a super-app. It truly is the Ride Hailing/Food delivery service app to use in certain parts of SE Asia. WeChat has a massive ecosystem in that its chat, payments, and a million company built "apps".
Not just Grab, most of tech giants in Asia (granted, I only know about the sinosphere) pursuit the "super app" idea. Here in Vietnam even the bank apps try to be super apps, it's somewhat hilarious but part of asian tech identity that makes us unique compare to the western tech scene.
The most puzzling one for me is AirAsia. Like I’m going to use an airline app to order food to my home. At least use a separate brand!
Banks all over the world do it . American banks do it too.
In many bank apps they will do travel, stay, cars, vacations, shopping and other dozen things like that .
It is not just banks , airlines do it too, they are so good at it, they are sometimes called finance companies with incidental air transport business.
Way off-topic, but is there a betting pool that Elon Musk is going to make it mandatory for the government agencies to use Super-X to do their work? And maybe for citizens too? It will obviously fail-whale all the time, but hey, he'll ignore the parts where it's not working when stroking his ego.
Somewhere there's a Dr. Strangelove kind of comedy where a wrongly addressed tweet sends the nukes flying, because the competent engineers on X will make an API over tweets, you see...
I’m guessing America no longer has the edge on high-speed pizza delivery.
Blinkit in India does 10 minute deliveries. It’s pretty mind blowing.
High population density + high Gini coefficient.
Still very logistically impressive. I believe they pack the bag as the customer is selecting items before full order submission.
I’m somewhat disappointed no-one spotted the Snow Crash reference.
Ha I didn’t. But thanks for reminding me about Hiro.
[dead]
There is lot of talk here about imaging, but it takes lot of work to turn bunch of random images into useful map. How are they doing that part?
A use of LLMs by search engines that I would like to see applied:
# At indexing time - The search engine use LLM to convert all the page content into a "common language" (English?!) before indexing.
- Use LLM to rank the page (detect page spammy/informational level)
# At query time - My query is translated to the "common language" - Result list most appropriate results unbiased by page language, but only by spammy/informational value.
Is this comment on the wrong article?
Yes, was about this https://news.ycombinator.com/item?id=42114990 . I don't know we ended up here! sorry!
Why would you like to see these things? It seems undesirable to return results in a language the querier does not understand.
It depends. Let's suppose we want to search info about the bio of historic people, for example Julius Caesar. Google will return for sure the link to english wikipedia. But let's assume that Italian Wikipedia got more deep info about Caesar. Would you prefer to see a less complete English version, or get the most complete Italian version (and auto-translate it in English with any available tools to read it)?
1 and 3 are done. (2) cannot be done practically by any dominant search engine because, no joke, the spam site owners get mad.