Learning-augmented robotic automation for real-world manufacturing

TL;DR

This paper presents a hybrid learning-augmented robotic system that automates manufacturing tasks with minimal data, demonstrating reliable, safe, and high-quality operation in real-world production environments.

Contribution

It introduces a novel hybrid system integrating learned controllers and safety monitors, successfully deploying it on an industrial line with limited data and achieving continuous, high-quality operation.

Findings

Operated continuously for over 5 hours with less than 20 minutes of real-world data per task.

Produced 108 motors with a 99.4% pass rate on quality control.

Maintained near-human takt time and reduced variability in quality and cycle time.

Abstract

Industrial robots are widely used in manufacturing, yet most manipulation still depends on fixed waypoint scripts that are brittle to environmental changes. Learning-based control offers a more adaptive alternative, but it remains unclear whether such methods, still mostly confined to laboratory demonstrations, can sustain hours of reliable operation, deliver consistent quality, and behave safely around people on a live production line. Here we present Learning-Augmented Robotic Automation, a hybrid system that integrates learned task controllers and a neural 3D safety monitor into conventional industrial workflows. We deployed the system on an electric-motor production line to automate deformable cable insertion and soldering under real manufacturing constraints, a step previously performed manually by human workers. With less than 20 min of real-world data per task, the system operated continuously for 5 h 10 min, producing 108 motors without physical fencing and achieving a 99.4% pass rate on product-level quality-control tests. It maintained near-human takt time while reducing variability in solder-joint quality and cycle time. These results establish a practical pathway for extending industrial automation with learning-based methods.