Our innovative approach, for an automative giant, introduces a cutting-edge anomaly detection model that uses sensor data from industrial robots in real-time.

By learning from healthy patterns, it detects irregular patterns signaling imminent malfunctions, enabling timely interventions. This predictive maintenance tool not only enhances reliability but also optimizes robot performance across the board.

The anomaly score also uses Dynamic Time Warping, comparing normal healthy cycles to the new tested cycle, as an added feature.

The solution was able to detect major malfunctions before they happened, allowing the manufacture to intervene quickly and solve the problem before a catastrophic breakdown of the robot.