Abstract
The design and implementation of a microcontroller vision-based manipulator control system is described. The control system is divided into high and low-level loops. As high-level devices computers or minicomputers have been used when conducting most research. In this work, a microcontroller board has been used as a high-level device due to its mobility and efficiency. A solution to the inverse kinematics task of a delta manipulator is presented in this work which forms the main part of the low-level device algorithm. Sipeed Maixduino and Arduino Mega dev boards have been used to implement the control system. The objects have been classified by YOLO model for the testing purposes of the control system. This has made it possible to measure the average targeting time for the different objects, which is used to test the implemented control system. The empirical data has shown a suboptimal result. The reason for it has been analyzed and addressed through coding a more optimal solution. Potential improvements to the hardware used in the test setup are suggested to benefit future work.