Inspiration
We all know whenever we are on Indian roads, traffic makes our lives most pathetic. It's the most underrated problem in India that needs to be solved right now. India still has those traffic lights which are given static duration at certain time zones in a day and region and there is also manual intervention and monitoring. It does not take into account real-time traffic behavior and hence result in severe congestion. This inspires us to build an autonomous traffic signals system which can optimize themselves to distribute the congestion in paths.
What it does
We provide a pluggable API that takes basic input parameters like Duration of a path with traffic and without traffic, default signal duration. Along with these inputs we tend to calculate probabilities of certain paths which are more likely to get traffic (this can only be done on data analysis)
How we built it
Here is how our algorithm work
- Data Acquisition: Calculating Min and Max time to cover the distance and probability of traffic movement in a certain direction using analytics.
- Find Max Value of Congestion Factor CF = (Tmax - Tmin) / Tmin
- Calculate Congestion capacities of neighboring paths.
- Finding Congestion movement per second
- Pick the most congested path and calculate the amount of congestion (ACi) that can be released into this path.
- Calculate the ACi that takes min time depending upon the path's congestion acceptance.
- Release traffic with this minimum amount of time.
Challenges we ran into
From the very start, this problem needs to be tackled first at the design level. There are hundreds of factors that influence the signal duration. The major problem is how to calculate them and prioritize them. Considering an example if we somehow calculate congestion density for each path in crossroad then how to distribute the congestion without making extra congestion in subsequent hops.
Accomplishments that we're proud of
We are able to build a basic algorithm which takes few considerations and can compute the real-time traffic signal duration and can even help to distribute the congestion. However, this mathematical algorithm takes some parameters that can only be found by analyzing data for a few months. We can say excitingly that we have solved (maybe 10% xD) of one of the world's most difficult problem in a very feasible manner.
What we learned
We learned how to tackle such difficult problems which require extensive logical thinking and powerful algorithms by using techniques like design thinking and validations. We have spent our 60% of the time in discussing and validating over numerous points and tried to build a tree.
What's next for Smart City
This system's capabilities are enormous! Our government if deploy this system it can handle congestion even over a single crossroad effectively. If used by the government at traffic signals it can collect such an important amount of data which would then can even optimize Google Maps.
Log in or sign up for Devpost to join the conversation.