Trajectory control based on On/Off, Fuzzy Logic and Convolutional Neural Networks for an Industrial Robot Arm: an experimental comparison

The objective of the present study is to compare three control approaches: ON/OFF control, fuzzy logic, and convolutional neural networks (CNN) implemented in Python for controlling the real-time trajectory tracking of a six-axis industrial robotic arm. This analysis has significant applications in fields that require a high level of precision, such as automated welding and surgical interventions in the medical domain. To evaluate the performance and adaptability of the control models, we will analyze the results using metrics such as Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), as well as metrics including Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), Jaccard Index, and Pearson's correlation coefficient. The results obtained reveal valuable information about the advantages and limitations of each control approach, highlighting the effectiveness of CNNs in visual perception and trajectory tracking. The ability of CNNs to interpret visual complexities is presented as a key factor for their success in industrial robotics and automation applications, suggesting a promising future for these technologies in dynamic environments.