Detecting Approaching Threats with the Police A.L.E.R.T
School: School of Dreams Academy
Area of Science: Computer Vision
The safety of our law enforcement is a problem that continues to grow in society. Just this year, an MIT officer was killed by two men in his vehicle. We depend on these people to maintain a safe civilization for us, but we cannot rely on their protection if they have none of their own. A simple solution to the issue is the Police A.L.E.R.T, a device that will sit on top of police cars and work to inform the driver of people approaching the car. The Police A.L.E.R.T will not only detect possible threats, it will also alarm the driver and give the placement of the threat in regards to the position of the car.
The goal of this project is to develop a vision processing algorithm that is effective and simple enough to be utilized with the Police A.L.E.R.T. This provides many challenges, as we are working with limited processing power, an unusual field of view, and a linux based operating system.
The solution to our problem will be found through the testing of different methods, the research of vision processing algorithms, and the development of our own program. We will need to gain a good understanding of C++ as well as how to effectively work with Linux.
OpenCV will be our main tool for this, running from a raspberrypi. By utilizing a conical mirror, a camera, and the raspberrypi, we will have a good platform to develop our code with. The reason for the conical mirror is to gain 360 degrees of vision. The camera we will use will be specific for the raspberrypi, and we will be running the Raspbian OS, a version of Debian, on the pi.
Progress Thus Far:
So far, we have developed a working proof of concept using a robot controller, a webcam, a motor, and a Christmas ornament as a prototype. The ornament works as a mirror for the webcam so that it has a full 360 degrees of vision. The program is written in the C language and can effectively track color reflected in the ornament and point in the direction it is coming from with the motor.
Our next step is to get our Raspberrypi working with OpenCV and the camera so that we can begin working on motion tracking with the pi itself. After that, we will work on the same concept except with an actual mirror. Once we have these things working properly, we will begin to alter our program so that it is more efficient and functional with the pi.
Our parameters for success include a functioning device, efficient code, and data from tests. At this point, we feel that the project will be quite successful and that we will have met our own parameters. If our project is completed in a timely manner, we will continue our research on vision processing and what kind of place it holds in the world of parallel processing.
Sponsoring Teacher: Creighton Edington
Mail the entire Team