AI's Gradual Rise on the Factory Floor

SAN JOSE, Calif. — Deep neural networks are steadily moving into industrial environments, poised to transform factory automation.

Early adopters already embed neural networks as the intelligence behind computer‑vision cameras. Over time, these networks will infiltrate robotic arms, sensor gateways, and controllers, reshaping manufacturing processes. The transition, however, is unfolding at a measured pace.

"We’re still in the early phases of what’s likely to be a multi‑decade era of advances and next‑generation machine learning algorithms, but I think we’ll see enormous progress in the next few years," said Rob High, chief technology officer for IBM Watson.

Neural networks are beginning to populate Linux‑capable, multicore x86 gateways and controllers that increasingly surround the factory floor. Emerging 5G cellular networks will eventually grant these systems seamless access to remote data centers, High added.

Automotive, aerospace, and healthcare sectors are already taking early steps, primarily with smart cameras. Canon is embedding Nvidia Jetson boards in its industrial cameras to activate deep learning. Cognex Corp., a leading industrial camera vendor, is expanding its own deep‑learning offerings. Chinese startup Horizon Robotics ships surveillance cameras that incorporate its deep‑learning inference accelerators.

"All the early adopters have deployed deep learning for visual perception, and others are starting to notice them," said Deepu Talla, general manager of autonomous machines at Nvidia. "Perception is reasonably easy to do, and researchers see it as a solved problem. Now the big problems are in using AI for interaction with humans and more detailed actuation — these are 10‑year research problems. In areas such as drone and robot navigation, we are more in the stage of prototypes."

Talla calls robotics "the intersection of computers and AI," but many industrial uses of deep learning will be less glamorous — and will arrive sooner.

Factory robots are not yet employing AI, says Doug Olsen, chief executive of Harmonic Drive LLC, a leading supplier of robotic components. In the short term, focus will shift from smart robotic arms to embedded systems that predict failures and collect usage data to schedule preventive maintenance. "That’s where AI can take hold first," Olsen explained.

Major chipmakers are aligning with this trend. Renesas began experimenting three years ago, deploying AI‑enabled microcontrollers at end nodes to detect faults and predict maintenance needs in production lines at one of its semiconductor fabs. In October, Renesas launched its first MCUs featuring dynamically reconfigurable processor blocks for real‑time image processing, with plans to release controllers supporting real‑time cognition in 2020 and incremental learning in 2022.

STMicroelectronics is taking a similar approach with its STM32 line. In February, it announced a deep‑learning system‑on‑chip and an accelerator under development, aimed in part at fault detection on the factory floor.

Smart robots will eventually arrive. Startup covariant.ai is working to enable them with reinforcement learning. "Equipping robots to see and act on what they see will be one of the biggest differences that deep learning will make in the next few years," said Pieter Abbeel, AI researcher, founder of covariant, and robotics lab director at UC Berkeley.

Abbeel showcases impressive simulations of robots learning to run using neural‑net techniques, but he cautions that the field is still in its infancy. "In fact, we started covariant in part because the industrial AI space is not that crowded yet," he said.