A snapshot from a session at Telco Cloud & Edge Forum this week, which asked about edge-oriented use cases in enterprises, and threw up a couple of interesting examples – which tell how computer vision, specifically, attached to edge-based cellular networks and computing systems, has emerged as a powerful application to workaday Industry 4.0 headaches. They were interesting because they were not about dreamy future scenarios, for once, but about prosaic contemporary holdups.
They were supplied by KORE Wireless and Volt Active Data, which argued that computer vision is being used in industrial setups to make legacy equipment ‘smart’, or at least to sidestep inherent ‘dumbness’ (for want of a better term), and to make smart equipment ‘dumb’ (ditto), and therefore to make it affordable. Both examples posit that computer vision – connecting live video on private 5G networks, nominally, to AI software in edge compute systems – can expedite Industry 4.0 by making it more accessible.
The first point, raised by Chris Francosky, chief information officer at KORE Wireless, is that industrial equipment is expensive, and upgrade cycles are long. The business case to buy IoT-enabled machinery, in order to run predictive maintenance and sync production schedules, is not an easy one. Computer vision, said Francosky, affords a way to fix an AI-enabled video feed on a machine or a process, and to get some of the same readouts. “You can’t embed telemetry in some devices, and AI video offers a way to observe what is going on,” said Francosky.
The second point, presented by Dheeraj Remella, chief product officer at Volt Active Data, is that the migration of compute workloads from the private edge to the public cloud over the past decade for service efficiency, and then back again more recently for performance and security has another twist: that compute power is also migrating, in some cases, from the far-edge, in industrial devices and machinery, to the private network edge (MEC). The purpose of this backwards trajectory is to unburden the machinery – to make it cheaper to buy and cheaper to replace.
Remella said manufacturers of autonomous mobile robots (AMRs), used in logistics and manufacturing to ferry goods about warehouses and factories, are producing simplified versions that attach to computer vision applications on private edge systems for their navigational ‘smarts’. “They want to move the intelligence from the robots to a near-edge layer so the orchestration is being done outside of the robots, and the robots become more commoditised. Because otherwise they are very specialised; replacement lead-times are very high,” he said.
All Telco Cloud & Edge Forum sessions are available on-demand here.
