Answers are grounded in this paper and cite their sources.
Due to the great rise of robots globally, such equipment has become fundamental for achieving a manufacturing plant's daily productivity and quality goals. A failure or anomaly that may occur with a machine can lead to a major stoppage of the machine or even the production line. In this context, the study of prior detection of anomalies in robots becomes suitable for planned preventive maintenance, production continuity, and equipment longevity, which is the main objective of this work. The methodology comprises constructing a dataset with positioning anomalies generated by a robot on a real production line. The Isolation Forest model was selected to operate as an anomaly detector. The model achieved a macro F1-score equal to 0.987, which is suitable for the monitoring in question, given that the model was developed in a real assembly line. This model was also implemented in the machine's controller with an execution response time of 565 ms, without compromising the controller's other tasks and allowing continuous assessment of robot failures—configuring the solution as an edge device. Therefore, this work extends the literature by demonstrating the detection of anomalies in a real production scenario, a subject underexplored in recent works.
No takes yet. Share an insight, caveat, or question.
Andrade et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: