Modelling traffic jams in major city junctions using Julia

[Prashant K V]

Recently, an article called The Physics Behind Traffic Jams, came up on the internet. It is a long essay on the dynamics of traffic jams in Seattle, after long observation. A little bit of traffic fluid dynamics. How about drafting a mathematical model of the same and putting up the test data on Julia? This will assist in determining traffic flow in important junctions during important days.

---

The Physics Behind Traffic Jams - SmartMotorist.com

[Prashant Venkatasubban]

Do you think we need data? We can start with modeling with parameters and constraints. Then we can scrap for data from traffic departments.

Cheers!
--

@prashantkv
Principal Organizer, Devthon

On Sun, Jun 23, 2013 at 2:08 PM, Bibek Padhy <bibekpa...@gmail.com> wrote:

great idea.. we can work on that.. and present it to google maps and city traffic police. We will need to collect some raw traffic data first for indian metros to shape the physics according to indian roads..

--
You received this message because you are subscribed to the Google Groups "Devthon" group.
To unsubscribe from this group and stop receiving emails from it, send an email to devthon+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Ravi Balgi]

Cool, do we have data that we can use to analyse.

As I could understand, traffic jam easing patterns are a problem of wave smoothing techniques.

mumbai traffic police beta tested a  service by fitting gps devices on taxis moving around the city and stating the traffic flow at important junctions in the city.

It is called Traffline, and is comprehensive enough.(they dont have interface though -http://www.traffline.com/)

Havent found data sets that have data about traffic count at any given time.

Here's what we can do, there is data available for tolls, which also map similar traffic behavior http://goldengatebridge.org/research/crossings_revenues.php .

There is a best chance to get data for toll crossings and create a model based on that.

Any thoughts?

--

Coherendz India Private Limited

M: +91 9869 23 5556

W: www.coherendz.com

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Build your Alumni Network, Visit www.vaave.com

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[Bibek Padhy]

As the author experimented and approximated patterns of the traffic flow in Seattel, we need to find out such patterns in indian traffic. 

The patterns we observe in indian traffic can be way different than what we observe in US.

For example what the author is trying to describe and solve is typical heavy traffic highway patterns which is common in U.S. but not so common in India. 

In india most of the traffic is dictated by narrow lane and abrupt road designs. 

It does not mean that we can't create a mathematical modle of the same but rather we need to come up with our own mathematical model which may not be the same as the pressure wave model but something different.

To find the model for our scenario we need to have data which we can use to reach to our approximate models.

We can start by gathering city specific data and generating mathematical approximation of the traffic in such areas.

[cestmoiaravind]

Assume, simulate and then validate with reality.

I don't think we would get the real data from existing sources unless we get real time pictures and count the vehicles.

Let's just take one junction. The Jubilee Checkpost in Hyderabad.

Towards Madhapur(left), Towards Yousufgouda(top), Towards Punjagutta(right kinda), and the other side.(bottom)

https://maps.google.co.in/maps?q=jubilee+checkpost&hl=en&ll=17.426654,78.416129&spn=0.003613,0.005563&sll=17.426654,78.416129&sspn=0.003613,0.005563&t=h&radius=0.22&hq=jubilee+checkpost&z=18

The parameters I see which would help

1. No of vehicles in a row in one direction, (at one level in the road, across different lanes)
2. The rate at which this traffic moves per road per direction
3. The average stop(red signal) time
4. The average go(green signal) time
5. The probability of a vehicle not complying with the go time (slower than the average traffic movement, hence causing a growing "wave" behind it)
6. Traffic density going towards each direction from each road

Please add more and correct me. I think this would be a good start to model. With assumptions.

The reasons why I choose this particular junction

1. There is a slope involved on one side, so the probability of driver mistakes should be higher than on a flat road
   
2. All 4 roads are of different widths and traffic density(to and fro)
3. Interconnected free lefts
4. I spend a lot of time these days stuck in the jam!