Oscar Hernandez

Oscar is an entrepreneurial engineer based out of Riverside, CA.

A Quick Glance at the Industrial Internet of Things

by Oscar Hernandez

Humanity, as we know it, has been shaped by the agricultural revolution, the first industrial revolution, the advent of mass production, and the imperialism of electricity, computers, robotics, and automation. Influenced by my friend and colleague Chris Hernandez, I am now especially interested in the fourth industrial revolution which is the current trend of automation and interconnection in manufacturing – read his article on the big problems in manufacturing. In this article, I’ll discuss three major solutions the Internet of Things has to offer revolutionary manufacturers.


The Internet of Things is the network of electronic devices that connect, interact, and exchange data across the internet. A typical manufacturer relies on the internet heavily to find and interview employees, set their schedule, pay them, record the facility’s performance and finances, identify raw materials, where and how to acquire them, how to compose them to build more valuable goods, who would want them, how to sell it to them, and evaluate their level of satisfaction. However, customers never have a perfect picture of what they truly need, manufacturers don’t understand exactly how their goods are being used, both of which cause grief to the engineers and designers in charge of improving their goods. Of course, smart phone manufacturers and mobile app developers record information including the frequency and duration of product’s usage, which allows them to A/B test new features. What can manufacturers learn from that?


The general strategy is to embed sensors in your equipment or goods to record all the data you deem relevant (or hope to be relevant), in order to identify patterns and anomalies in real time before things break. You could embed a sensor on a motor to identify when it’s not operating at an acceptable velocity, so that you can replace it before it breaks eveything else. You could embed timers on goods to alert the end user when your good or warranty expires to keep them safe, and to alert you to check up on them. You could also operate your facility remotely if it’s connected to the internet, like unlocking the doors for your staff while you’re away at a conference, or halting production from a safe distance if the STOP buttons are inaccessible or damaged. You could add power sensors to your building to check that you’re not overpaying your utility bills. To be dogmatic, collect all the data you need. Collecting all this data, however, is only useful if you can make sense of it.


Data is context, and context becomes more valuable with more relevant context. Identifying anomalies in a part is good, diagnosing failure in a complex system is better, and fixing it automatically is ideal. This is more feasible now than ever because of recent developments in big data processing, cloud computing, applied mathematics, data science, artificial intelligence (AI) and machine learning. The general solution is to identify a failing system, identify which parts are malfunctioning and how long they’ve been malfunctioning, and order its replacement while you minimize the risk by decreasing the system’s production rate or hibernating the system. On a car, this means using all of your sensors (tire pressure, check engine, oil, odometer) to address the issue quickly instead of having to wait for your car to break down to recognize the issue. For a facility with a battery management system and an unusually large energy bill, this means feeding the relevant data to an AI algorithm to control when to charge/discharge the battery. Often, a simple camera is a great sensor – for a primer, I recommend Rany Tith’s article on Computer Vision. All that being said, data is often sparse or unintelligble, and it’s very difficult to gain valuable understanding from it; that’s why fields like AI and data science are popular and important. Now that we can diagnose and prevent pitfalls, you should aim to manage your processes to optimize for resource usage, including human labor, cost, energy, and time.


Optimizing your processes is, of course, also very difficult – I recommend Xavier Hernandez’s article on fully automated warehouses. Rudimentarily, this means programming robots to do stuff and teaching humans how to work around it. There are also developments in programming human-and-robots distributed operating systems, so that robots account for and collaborate with humans. I’m excited for humans to stop doing robotic things, and start doing human things. I hate clocking in my hours – I wish I could just download a concise human-readable log of my computer history (beyond browser history), and I’m sure manufacturing operators around the world would rather be recognized when they enter the facility and skip the medieval punch cards. I hate driving cars and I’m sure I’d hate driving a forklift, so I’m pushing for the advent of autonomous vehicles that drive themselves. I hate reading long-winded manuals – I, like many, would rather strap on friendly augmented reality goggles to teach me what I need to do.

Personal Call to Action

As an ambitious computer engineer, I get to learn bleeding-edge network theory, program robots’ operating systems, build web apps enabling internet communication with humans and machines, and partner with other mechanical, systems, and software engineers to bring them to life. It’s a brave new world, and I want to help manufacturers taste the fruits of IoT, AI, and AR.