The most significant risk next few week is not being able to implement a properly working crash detection algorithm. This is simply because this is probably the most challenging part of our project to get working. It also serves as a kind of roadblock as the future components of our project rely on getting this one working well. Thus, we would like to spend a lot of time next week on this. Our contingency plan would be to move forward with an algorithm that may not be working super well, but working well enough to implement future components, and then go back and try to retrain / redo the initial crash detection algorithm. Other than that, we have done a lot of research on the various components of our project, such as the hardware to purchase, the data and neural network / algorithms to use to train for crash detection, and formal / algorithmic ways to think about traffic rerouting. We definitely need to spend some time collecting our thoughts and start implementing what we have in mind. There are no changes to the schedule / future plans.
Arvind’s Status Post – 02/12/22
We are currently in the research / gathering information stage of our project. I have found a research paper that appears to follow a similar process to the one we have planned, where they use computer vision algorithms to do vehicle detection and then track the vehicles’ speeds and positions to determine crashes. This paper could be useful in addition to the data / information my team mates have gathered.
We will need to simulate traffic environments with real hardware. I found this project here: https://create.arduino.cc/projecthub/umpheki/control-your-arduino-from-your-laptop-via-wifi-with-esp13-346702?ref=part&ref_id=8233&offset=25
This project involve an arduino board with a shield that allows it to connect to WiFi. It also uses the wifi communication to control an LED. This is very similar to what we want to simulate, as we will be using LEDs as our “traffic lights.” We think WiFi is the best wireless communication protocol to use, as we will have to communicate from a computer to our Arduino boards as the traffic detection is being run on a remote computer. WiFi makes this connection easy. It also simulates how our system would be potentially implemented in real life, where a traffic light would have to wirelessly communicate with another traffic light farther away. I think purchasing the components of this project and building it would be a good place to start. We can then make modification to suit our purposes better and experiment.
I think we are on track in terms of schedule but do need to start collecting and implementing our ideas. By next week I hope to have started and potentially finished constructing the above project depending on how quickly we can purchase parts.
Goran’s Status Report – 02/12/22
For this week our groups focus was mostly on researching different types of algorithms/papers that we thought would be helpful to the implementation of our Smart Traffic Light. My focus for this week was researching dynamic traffic rerouting algorithms. The idea behind this was in the case of some sort of accident result in lane closures or congestion, we would be able to control traffic lights at nearby intersections in order to ease traffic slowdown. From these papers I believe I have gained a reasonable grasp on the next steps I would like to take for the following few weeks. I will take several different example road maps (different configurations of connected intersections), and convert these into graphs with nodes and edges where the edges represent the roads and the nodes represent the various road intersections. The roads are given various weights depending on their distance and traffic density, and the time a motorist would take to move from one point to another is estimated. These weights would also be heavily dependent on data coming from the crash detection algorithm, allowing us to factor the effect of crashes on our traffic simulation. I will then add in a suitable routing algorithm (most likely just Dijkstra’s) and apply that to the weighted graph. In the future I plan to get this rerouting set up to simulate on a series of breadboards with leds.
Right now I feel as if I am on schedule, the amount of research I have conducted this week feels like enough to begin the implementation section of our schedule. Hopefully the entire rerouting algorithm can be completed at/before the three week deadline as seen on the schedule.
By next week I hope to be able to show reasonable progress on the rerouting algorithm, and at least have it working on a preset input of weights on one road map configuration.