Camera System for an Automatic Placement Machine
The project consists of a sturdy structure that can be moved as a single unit. The camera uses a Raspberry Pi to calculate the centers of the components in real time and can provide both this information and a live feed through which these centers and their positions can be viewed.

Camera System for an Automatic Placement Machine
Degree Program
ET - Discontinued
Project Description
The project consists of a sturdy structure that can be moved as a whole. The camera uses a Raspberry Pi to calculate the centers of the components in real time and can provide this information, as well as a live feed from which these centers and their positions can be retrieved.
The main objective of the project is to detect the centers of SMD resistors up to size 0603 (inches) in bulk using a camera. The setup is designed to allow for easy, user-friendly configuration and data retrieval. The lighting assists the camera in determining the component centers.
People Involved
Student group: Sebastian Summer, Manuel Melcher, Tobias Ruisinger, Michael Bablok
Advisor: Prof. Dr.-Ing. Christoph Zeuke
Course
Systems Engineering 1 (IK)
Duration
Winter Semester 2023/24
Requirements
Hardware:
- Sturdy construction with maximum dimensions of 80 cm x 70 cm x 70 cm (WxHxD)
- dimmable lighting for the bulk material
- Camera height easily adjustable without tools
Software:
- Web server with access to a live feed and position data
- Marking of component centers
- Calibration of lens distortion and standardization of pixel width
Implementation
The structure consists of
- item profiles,
- 3D-printed mounts,
- a metal slide with fine adjustment, and
- a base that can be easily removed.
- A Raspberry Pi HQ camera
- with a Pentax C-mount 35mm lens,
- is processed by a Raspberry Pi, which handles component recognition.
It also sets up a local web server through which a live feed can be accessed and started, a JSON file containing the position data of recognized components is provided, and images can be downloaded.
The lighting can also be controlled, and calibration can be performed for the set camera height, which remains valid for subsequent images taken with the same camera settings. The latter also enables the standardization of pixel width and thus position by reading a known grid pattern.
The software is written in Python, and the main image processing library is OpenCV.
Uses
The data for the position can be transferred to another project for an automated placement machine, such as one with a suction cup arm.
Of course, when dealing with bulk material, the quantity can also be easily determined and used as a counter.
Integrating OCR would also enable component value recognition; the raw data in the Pi would allow for this, though the Pi itself lacks the processing power to handle it.












