Ever-Smarter Robots Struggle to Explain Themselves
The deployment of artificial intelligence in industry continues at a breakneck pace. Industry is adopting state-of-the-art AI techniques, like reinforcement learning, and neural network architectures, such as Transformers and LSTMs, with great results (see Ford below). But how to practically augment an industrial AI system with explainability is an open topic. Current explainable AI (XAI) techniques such as Shapley values can be traced back to the 1950s and effectively treat AI as a black box. Eventually, the bottleneck of using AI in manufacturing environments will be due to the lack of explainability in AI systems.
Visual Inspection
Microfactories in Action
Acoustic Monitoring
Assembly Line
Start-ups Powering New Era of Industrial Robotics
Much of the bottleneck to achieving automation in manufacturing relates to limitations in the current programming model of industrial robotics. Programming is done in languages proprietary to each robotic hardware OEM – languages “straight from the 80s” as one industry executive put it.
There are a limited number of specialists who are proficient in these languages. Given the rarity of the expertise involved, as well as the time it takes to program a robot, robotics application development typically costs three times as much as the hardware for a given installation.
Application Layer Protocol Options for M2M and IoT Functionality
With adoption of Internet of Things (IoT) and Industry 4.0 functions, devices are increasingly connected via industrial protocols. What’s more, today’s machine to machine (M2M) communications are rapidly standardizing on these protocols. Complicating matters is that IoT protocols don’t describe a single application-layer protocol, as several standards are in operation. So while early IoT implementations used standard internet protocols, there are also dedicated IoT protocols now available.
Modeling communication structures and identifying the right protocol for a particular application can be daunting. This article outlines what various protocols do as well as the options available for these protocols — so design engineers can more easily select the most suitable to integrate.
Ford's Ever-Smarter Robots Are Speeding Up the Assembly Line
At a Ford Transmission Plant in Livonia, Michigan, the station where robots help assemble torque converters now includes a system that uses AI to learn from previous attempts how to wiggle the pieces into place most efficiently. Inside a large safety cage, robot arms wheel around grasping circular pieces of metal, each about the diameter of a dinner plate, from a conveyor and slot them together.
The technology allows this part of the assembly line to run 15 percent faster, a significant improvement in automotive manufacturing where thin profit margins depend heavily on manufacturing efficiencies.
How Pfizer Makes Its Covid-19 Vaccine
“This is where the magic happens.”– Patrick McEvoysenior director of operations and engineering
A rack of 16 pumps precisely controls the flow of the mRNA and lipid solutions, then mixes them together to create lipid nanoparticles.
When the lipids come into contact with the naked strands of mRNA, electric charge pulls them together in a nanosecond. The mRNA is enveloped in several layers of lipids, forming an oily, protective vaccine particle.
Synchronizing eight pairs of pumps is not an ideal solution, but Pfizer engineers chose to scale up existing technology instead of trying to build a larger, unproven type of precision mixing device.
The newly made vaccine is filtered to remove the ethanol, concentrated and filtered again to remove any impurities, and finally sterilized.
OPC-UA: the Universal Language of Industry 4.0
Forgive the obscene title of this article, for implying OPC-UA is nothing but a simple communication protocol is a great injustice. Indeed, OPC-UA encompasses this, but also so much more. It is a living, breathing, specification: one that outlines an information-centric architecture that is interwoven with security systems-systems which permeate a definitive rule-set for device modelling and communication.
At its essence, OPC-UA is a platform-independent, machine-to-machine communications architecture that focuses on providing an object-oriented approach to modelling data.