Week 7

02/19/2024 - 02/26/2024

This week consisted of us deciding the best way to implement the pet detection system to operate the door. The three options that were being considered were a pressure sensing mat, an RFID reader and tag, and a camera to recognize when the pet is in proximity.

When considering the pressure mat, the following issues were noted:

• Inability to distinguish between a pet or other animal.

• Bulky design that sits in front of the door.

When considering the RFID reader, the following issues were noted:

• Lack of range for the RFID reader.

• Longer range readers are expensive and impractical.

We chose to use a camera interface that allowed the door to open when the pet was detected. Cameras are compact and can be mounted in a variety of configurations to comply with the project specifications. They are also inexpensive and easy to obtain. Using the camera interface instead of the RFID reader or pressure mat will remove all of the above noted issues. The Raspberry Pi will be running the TensorFlow object detection library to allow the cameras to detect when the pet is within a pre determined range and open the door accordingly. TensorFlow is a large software library that is able to use AI to determine the objects in a given image and their location in relation to the camera. This library was chosen due to it it being much faster and using less resources than other libraries, making it perfect for a Raspberry Pi applications.

In addition to deciding the means of pet detection our group also spent time in the open lab analyzing the door motor on the doggie door that was purchased for the project. The first step was testing the functionality of the door, then disassembling the door itself and analyzing each of the components. Once satisfied, we connected the wires from the motor to the oscilloscope and noted the output waveform. Upon pressing the "open" button, the voltage output went from 166mV to 3.84V DC. Once fully open, the voltage returned to the resting state. Similar results were observed when pressing the "close" button except the polarity of the voltage was changed. An output voltage of -3.44V was recorded at this time. Recorded results are shown in the video below.