You’re new to the city and plan to go to a movie theatre on a Saturday evening. You have no idea where this theatre is because chances are you’ve booked the tickets online. So you do what anyone with a computer and an internet connection would do. That’s right ...Google Maps. You go through the drill ... point A, point B, get directions and you have it, the shortest path, the distance and expected time right in front of you. There’s just one problem, I think in our cities we need to start measuring how far places are in terms of time rather than distance. But doing that conversion requires knowledge of way too many parameters like time, day of the week, road conditions, traffic patterns, weather, construction projects and even events like political rallies, festivals etc all of which vary from place to place.
Well the problem and the causes are pretty obvious, what we need is a cheap and effective solution, cheap being the keyword. One thing that is common between every street in every city is the ubiquitous TAXI. Recently, in India, we have seen private companies entering this business. The interesting thing is most of the taxis owned by these companies are equipped with a GPS. Basically, if you look at it another way, you’ve got thousands of these GPSs crawling all over the city’s roads 24*7 and that is going to generate a lot of useful data.
Here is how it can work, all the data from the GPS’s is uploaded to a main server. This effectively gives you a snapshot of the city map with lots of dots corresponding to positions of all the taxis. Now let’s take these ‘snapshots’ at regular intervals (let’s say every 2 minutes). Comparing successive snapshots, find how much every dot on the map (or taxis) moves in the 2 minutes.
Then we use that wonderful equation that we all learned in 2nd grade. Speed = distance/time. That’s it!!
Here is a sample of a simple city and two taxis that gives us the instantaneous speed for two streets
At 7:00 pm At 7:02 pm
Do this for the whole city and you now have the average speed of traffic on every city road at that instant. Now if you do this, let’s say at 7pm on a Wednesday then it’s pretty obvious that these values of speed would hold good every Wednesday at 7pm.
That brings me to the next part, if the above calculations are repeated every 2 minutes for an entire year; you would get a good idea of variation in average speed over the year (for e.g, streets in low lying areas would show a drastic reduction in the rainy season). That takes weather out of the equation. 26th July 2010 will have the same weather pattern as 26th July 2011. Basically the longer you observe, more patterns start showing up.
One instant: road conditions
One day: rush hour traffic pattern
One week: daily pattern i.e. weekends vs weekdays
One year: weather patterns
Additional features
A graphical representation of avg. speed data would show congested roads (low avg speed) in red, clear ones in green..etc. Once you have a good database of traffic patters, say after 2 years or so, a time varying traffic map can be made. You drag a pointer on a timeline and the map varies accordingly.
Once, this is achieved, the applications could be huge. Other than obvious ones like traffic management and city planning etc there are environmental benefits as well.
Down to business
Isn't it always about the money? More often than not, for something to work, someone needs to make a lot of money out of it. Who pays for the server setup, the software applications, system maintenance, the GPS etc?
An interested group would be real estate developers (to do a quick analysis of how convenient a place is). Railways could adjust train timings to suit traffic flow. In the future, if we are able to make smart cars, their developers would certainly be interested in data like this. (I'm allowed one lame example). But, imagine a car that downloads data from one of these servers about the route with the best traffic flow and then drives you through it.
On a more serious note, here's one scenario, cab companies agree to share GPS data with let’s say, Google for a hefty fee. Google then uses this data in its maps application to give you a very good estimate of TIME taken to go from your home to the theatre based on data collected on a Saturday evening last July and you find out, the “shortest” route is not that short after all.
What’s in it for Google?
It adds relevance to Google map’s results as they now take into consideration local factors
What’s in it for the cab companies?
Money out of nothing!
Posts
Well, Aditya that's a really good idea there.
ReplyDeleteBut successful companies like Google would rather take information from a third party source rather doing the first hand hard work.And I guess they already obtained the data for some cities since Google Maps do predict traffic for cities like Los Angeles, New York, Chicago.
And the idea would not work in India as majority of traffic here is due to rather cheap cars and auto which obviously don't have a GPS system.
Though there are a lot of taxis with GPS system they rarely move because Indians (being cheap) would rather prefer an auto than a costly taxi.Hence it would take years to obtain the data.
Also as the metro's slowly creep into our cities and the "Going Green" awareness of the people are increasing people would prefer to opt public modes of transport.
That's the advantage of this solution, you don't have to observe the entire traffic because the speed of one car on a street is almost the same as any other. So, all you need is one GPS enabled taxi for an entire street.
ReplyDeleteAbout the number of such taxis, I checked out figures for one of the local cab companies, they own almost 2000 cabs in every major city in India and they are expanding, so obviously business isn't bad.
And about people preferring autos over taxis, a strange thing about Mumbai is, in certain areas auto-rickshaws are not allowed to operate, there are only cabs, and these are the areas with all the traffic jams.
But your point still stands, a setup like this will take more than 2 years to produce usable results.
Idea is gud only but I am not sure if cab companies are the right parties to do bussiness with.Like dinesh said, cabs mite b ubiquitous in newyork but if u r taking abt India, u need to partner with the likes of BEST or APSRTC. Maybe they can operate on a public private partnership like BOT. google provides the necessary hardware n the bus companies give data to google. The bus companies could also use this to have cool hitec bus stands where they have GPS dots pointing the exact locations of the buses so that commuters ll hav a better idea of how long they ll hav to wait..
ReplyDeletesay if this system is implemented,and on a saturday it predicts that some street has less traffic.so,ppl will start using that route...how many weekends shud go by before ppl can realise that now the same street is getting crowded bcos everyone is using it?
ReplyDeletethe areas with less traffic might be long (which less ppl might prefer after a long day's work)and this long route might contain a small part of the busy traffic area of some other route...
that'll make things worse.
so,its better if google finds out some formula(or,function :P) considering the long path and the traffic conditions occuring in between thus, showing the google 'like'ability of a route on scale of 10 or something...(this is my lame example)
ohh...sry!! i 4got!!...nice idea...nice idea shetty bhai!
ReplyDeletelol another factor mite be the popularity of this feature. If this predicts less traffic on road X and more on road Y,this itself mite be a reason for a substantial no of ppl to choose X over Y and this could result in the traffic of Y becoming less than X...but this ll prolly happen over time and not immediately..
ReplyDeleteFirstly, Aditya change the time standard of ur blog..
ReplyDeleteI posted my comment at 5:30pm on 6th or so.. not 4:25 am..
And nice point by Aravind...He shrewdly pointed out that the dynamic nature of people wasn't taken into consideration here. With people being really familiar with internet these days and with social networking sites on the roll I doubt that it would take long for the feature to be popular (i.e. if its released).In that case for the feature to work efficiently a dynamic solution (preferably with Artificial Intelligence) is required.
@pradeep
ReplyDeleteYes, it would be great if our buses could have GPS but there are 2 issues with that
1. Setup cost, with BEST already not having financial problems, it could be difficult to convince them to spend on something like this. In think they do have this feature in metros in the UK but I don't think we can afford it.
2. Buses are not a good reflection of traffic flow, because they don't travel point to point like cabs, so they would underestimate avg speed.
@aravind : Should have thought of that! This is not a problem though because the whole point of this system is to estimate and not improve traffic flow (that needs infrastructure development), the best it can do is even it out which as {Pradeep mentioned is going to happen in the long term
Setup cost is coming from googles pocket na...y ll best hav to worry...but ya, buses wont be a gud reflection of avg speed...
ReplyDelete