|
Named after muhammad al-khorezmi tashkent university of information technologies fergana branchBog'liq m12Stepper motor: Stepper motors are gaining ground in the electronics world. From a simple surveillance camera to complex CNC machines/robots, these Stepper Motors are used as actuators everywhere as they provide precise control.
Pic. 2.2.8. Stepper motor.
You can usually control stepper motors in different modes, and in addition, they have a certain transmission ratio. Each step corresponds to a rotation of 11.25 degrees according to the data. This corresponds to 32 steps per rotation
(360 ° / 11.25 ° = 32). In addition, the manufacturer specified a ratio of 64: 1 for this type of engine. Usually this gear ratio should be multiplied by 32 steps. If you study the 28BYJ-48(a type of stepper motor) in more detail, it becomes clear that the actual gear ratio is about 63.68395: 1. Therefore, we set the final number of steps to 2038 (32 * 63.68395 = 2037.8864) [31 ].
Pic. 2.2.8. Stepper motor connection diagram.
Servo Motor: Servo motors are wonderful devices that can be returned to a specific position.
Usually, they have a servo arm that rotates 180 degrees. Using the Arduino, we tell the servo to go to a specific location and it goes there.
Servo motors were first used in the remote control (RC) world, usually to control RC cars or flaps on an RC airplane. Over time, they found their place of use in the world of robotics, automation and, of course, Arduino.
Pic. 2.2.9. Servo motor.
It works as follows. If the shaft is at the right angle, then the motor will turn off. If the circuit detects that the angle is not correct, it rotates the motor until it is at the correct angle. The output shafts of the servo are able to travel around 180 degrees. Usually, it is in the range of 210 degrees , but it varies depending on the manufacturer. A simple servo is used to control angular motion from 0 to 180 degrees[28].
Pic. 2.2.10. Servo motor connection diagram.
Pulley: High quality GT2 pulley with 60 teeth and 8mm inner diameter. This pulley fits the GT2 9mm timing belt and is used to guide the timing belt smoothly.
Pic. 2.2.11. GT2 60 and 20 tooth pulley.
Drive pulleys can be fixed to a shaft or bar, unlike drive pulleys, which have a bearing that allows them to rotate freely. Most drive pulleys can be secured using two set screws, resulting in reliable adjustment. To drive the belt, we offer two types of pulleys: toothless and toothless (smooth). Wheel pulleys allow you to direct the teeth of the belt without the vibrations caused by working on a smooth surface[31].
The general idea is to use an Rpi camera to take photos with a certain frequency, process the image, detect a car number plate, recognize characters and compare it with the list of allowed numbers in a database. If it matches with the number plates in the list, the barrier will open.
At a baseline stage we will use the following tools:
As an image source — Raspberry Pi Camera module v2;
Numberplate detector — Yolo v7 served using pyTorch;
Optical Character Recognition (OCR) — EasyOCR;
"Database" — table in a Google sheet;
All processing jobs and calculations should be executed locally on Raspberry Pi 4b, the solution must work autonomously.
Pic. 2.2.12. Simplified diagram for baseline version.
Raspberry Pi will continuously read frames from Pi camera in a "near real-time". Then fine-tuned on a custom dataset YOLOv7 model will detect region with a number plate. After this, if necessary, provide image preprocessing and then the EasyOCR model will detect the number from the provided cropped frame. Then the number plate string is checked whether it matches any of the stored number plates in the "database" and the corresponding action is executed. Using Raspberry GPIO (General-Purpose Input-Output) — controlled relay switch we can connect the parking barrier and any additional load like lights, etc. GPIO pins also allow to connect input sensors (like IR, PIR) and trigger the camera only when the car is detected.
O nce again, this task can be solved in many ways, maybe some of them will be more efficient and easier for certain requirements and use cases. For example, all heavy processing could be performed in the cloud; we can use GPU-based edge devices, we can use other models; serve using ONNX, TFLite, etc. But this project was done as an experiment, using equipment that I currently have, and I was not looking for an easy way.
Pic.2.2.13. Raspberry Pi with a camera module
|
| |
