The Circular Audio Detection System, or C.A.D.S., is a microcomputer project for CSCE 462 during the Fall 2013 semester at Texas A&M University and was inspired by military-grade sniper detection systems. When powered on, the system will actively listen for and loud noises such as claps. When a clap is detected, the system will use the four mounted microphones to triangulate the direction from which the sound came from. Once the direction is calculated, C.A.D.S will point its mounted camera toward that direction and send a signal to a computer running our software. The software displays both the screenshot and live feed from the webcam.
The core concept of our project is using sound triangulation to locate the source of a sound. We envisioned using four microphones in a cross configuration as detectors. Using a timer to count the intervals of the sound wave’s propagation to the microphones, we can detect the direction of sound by triangulation. This is feasible because sound travels relativity.
Our overall goal for this project is to expand our knowledge on microcontrollers and hardware / software integration by implementing a directional acoustic sound detection system. The individual goals we are trying to meet are as follows: – Construct a frame which will house the PSoC, microphones, wiring, and camera / motor assembly – Develop a software approach to acoustic triangulation – Using the calculated location, interface with a motor in order to turn the motor to the direction of the sound – Obtain decent accuracy, such as no more than 30° deviation from the actual location of the sound.
Our project used the PSoC 5 Development Kit and Adurino Uno as microcontrollers for the microphones and servo motors respectively. Using PVC piping, we constructed a frame to hold the microphones in the four cardinal directions. The base houses the PSoC, Adurino, necessary breakdowns for wiring, and PVC pipes for stability. The top houses a camera mounted on two servo motors which turns towards the direction of the sound. A laptop is then signaled to capture an image of the sound source.
One issue that we ran into that was a big show stopper was getting correct data from the timer. There were not a lot of resources online, and what we had was not very robust. After many, many hours of tinkering, we finally we able to get counter data from the PSoC that made sense.
Another issue we ran into was power. Different power sources and different loads lead to different results. The PSoC was really finicky. Different voltage values and different components (such as the two different ADCs) lead to different threshold values. This made trying to get sane results difficult, because it was like chasing down a moving target.
Communicating the degree information to the servo controller also took some tinkering. This was because the voltage values were not exact to the specifications. We had to rip out the multimeter and do some math to find the correct milivolts per degree. The voltages were not very stable either, especially when running on certain power sources, so we had to edit the servo controller software to ensure the camera was not freaking out every second.