Intelligent ambient lighting for image processing based object recognition in robotic systems

In image processing-based applications, ambient lighting is an important issue in terms of accurate detection of objects and energy use. Ambient lighting in classical methods; It is done by keeping the light source on all the time, turning it on and off manually as needed, using a timer or automatically measuring the illuminance level with the help of various sensors. In this thesis, an embedded system designed for reference-color-based intelligent ambient lighting based on image processing for object detection and position determination for robotic handling applications has been designed and its performance has been examined. In order to obtain the minimum illuminance level in the system, the power of the lighting system is adjusted by reference to red, green, blue and yellow objects whose positions are determined beforehand, without the need for an external sensor. Raspberry Pi 4B mini computer and Arduino CNC shield board with GRBL software is used for image processing, lighting and 3-axis handling processes in the designed system. Image acquisition was done via the Raspberry Pi camera module, and the handling of objects was done with an electromagnet. For lighting, 12V DC white strip LED lamp driven by PWM signal was used and smart ambient lighting was made by controlling the illuminance level proportionally. System interface software is coded in Python language and OpenCV library is used for image processing and object detection and position determination. Within the scope of experimental studies to examine the performance of the designed system, white light was illuminated, a non-reflective white matte plastic background was used for the platform floor, and the platform was covered with a black cover. In object detection with color-based image processing, the minimum required platform illuminance values were investigated by using reference objects with different colors such as Red, Green, Blue and Yellow. According to the experimental results obtained; it has been understood that color selection is important in the detection process and it has been determined that the color that needs the least illumination level among these colors is yellow. It was observed that yellow color was followed by green, red and blue colors, respectively. It has been understood that in the designed system, color-based object detection can be done successfully at an illuminance level of about 50 lx and a power ranging from 0 to 225 mW is needed depending on the environmental lighting situation. In conclusion; It has been observed that the embedded system, which makes smart ambient lighting, provides the minimum illumination level with energy saving that changes depending on environmental conditions.