As a part of my Final Degree Project, and with the help of one of my colleges, we develop a system to detect people on top of a sensing mat, with the idea of offering an alternative to camera people-counting systems.
The system was composed of two parts, the sensing mat, with capacitive sensors that connected to the part I developed, a Raspberry pi in charge of the data collection and analysis.
The sensing mat consisted on two metal plates (20cm x 20cm) separated using a thin layer of felt. When someone step on top of the sensor, the felt compresses and the capacitance between the plates increases, being very easy to detect by a micro-controller by measuring the rise time of a square signal. The data was then sent to the raspberry through CAN bus.
To provide the raspberry pi with a bus interface, I designed a can shield, with a can driver and a RTC, all perfectly integrated seamlessly with the Linux kernel.
The first prototypes were made at home with the 3D printer.
The analysis was made using OpenCV, upscaling the 2D image taken from the sensors and performing a very basic algorithm for object tracking, with this, we were able to track the persons in top of the mat.
No hay comentarios:
Publicar un comentario