When Mary Shelly wrote Frankenstein in 1818, little could humanity imagine about the fourth industrial revolution nowadays. The Industrial Revolution in the XIX century fundamentally changed economic activities, gender and power relationships, and social classes. Not only could we manufacture products faster and more safely but also with less effort. Introducing machines, chemical reactions, and new power sources in factories facilitated our lives in many ways, almost as much as the digital technologies and the Internet today.
When the Internet was booming in the 90’s, we were convinced that all sci-fi movies would become a reality by 2000. Even though we do have video calls, cleaning robots, and e-commerce, we are still implementing and developing Industry 4.0 technology. The Internet of Things is one of those recent trends that many companies are trying to include in their digital transformation strategies to be able to holistically monitor and manage their facilities, factories, data centers, stores, etc.
What happens when IOT meets robotics?
Commonly the Internet of Things can be defined as the combination of hardware, sensors/actuators, infrastructure (middleware, Cloud, business processes, big data), networks/internet, and software. In the past 10 years, big business players worldwide have been leveraging IOT technologies to gain enterprise-wide visibility of their operations for preventive and proactive monitoring and manufacturing. IDC for example estimates that by 2025 there will be 55.7 B connected devices worldwide.
As explained in a recent blog, since Ancient Greece we have been trying to outsource dangerous, repetitive tasks to machines and we have been successful in building robots since the XX century. In the same blog we explained the fight for resources within robots and how designers and developers must carefully choose processing capabilities (Here you can continue reading about his subject).
Currently, many robotic projects include Artificial Intelligence, Deep Learning, and Pattern Recognition. This means that they gather, manage, and act based upon large quantities of data. Since much of the processing cannot happen inside the main robot infrastructure due to the limited resources within, many new robotic projects leverage the remote brain architecture. In other words, the main robotic brain is hosted in a server within a data center or rack while the robot’s body has a smaller processor just to gather data and execute actions.
These data intensive robotic applications face unique challenges in terms of hardware. Below you will find some of the most common.
Compatibility
Internet of Robotics Things requires companies to work with at least 5 layers simultaneously: hardware, software, network, infrastructure, and sensors/actuators. Unfortunately, not all these technologies were developed to work together, and many vendors lack brand agnostic equipment that can work with several communication protocols or programming languages.
In fact, if you are already working on factory automation, you are aware of the communication protocol nightmare. Attempting to connect sensors and actuators to PLCs and, later, to servers can become a costly and resource intensive enterprise. Luckily, the industry is developing brand agnostic standards for hardware OEMs to produce equipment that can translate and work with several communication protocols. You can read more on this blog.
Moreover, if you are leveraging 5G as well as the Internet in your robotic application, you probably need to consider the compatibility of additional modules. Some Hardware OEMs can help you integrate PCIe and LTE modules in the factory, avoiding you installation and operation headaches down the road.
Specialized Operating System
Robotic applications require a specialized operating system. The most current and successful nowadays is called ROS. This OS is fully customizable which allows remote processing and parallel programming, and it includes algorithms, libraries, and API management. These features are the foundation of remote-brain robots. This type of robots usually works with several programming modules and languages at the same time which means that you probably have a team of engineers working together. It also means that you store libraries remotely while processing data.
Since ROS cannot provide a list of approved hardware capable of handling the OS, you should find a Hardware OEM who can assess your needs, offer a customized solution and test that ROS runs smoothly. You can read more about ROS in this blog.
Robust, powerful, fanless servers
Remote brains in robots entail a specific set of requirements. First, processors in robotic servers need to handle large amounts of data. Therefore, developers would regularly use the latest and most advanced processors. However, you should be cautious about oversizing components. Partner with a Hardware OEM that can help you accurately assess processing capabilities.
Durable servers include as few movable parts as possible to reduce the MTBF (Mean Time Between Failures). As a result, you would probably choose a fanless server which will prove to be a great investment to reduce maintenance and repair costs. This feature should be paired with a solid-state hard drive and secure cable connectors.
Second, if the server will not be hosted within the controlled environment of a data center, you should probably assess if the rack offers the right ingress protection against humidity, water, or dust. If you are leveraging Edge Computing racks, you should inquire with the Hardware OEM if the server you are choosing is prepared to handle the harsh conditions at the production floor or outside.
Security
When you connect devices to any network you are opening the door to possible security threats. If you implement the Internet of Robotic Things carefully the advantages will outweigh the security concerns. To guarantee that you do so, you need to pay special attention to the hardware and the network protection layers.
Remote brain robots are considered distributed infrastructure. This means that you have hardware installed outside the secure data center environment which makes it vulnerable to unauthorized access and tampering. You can read about the common hardware security threats and how to avoid them in this blog. In summary, you need to look for untapped security features within hardware and consultation with the OEM to add security hardware as necessary.
Regarding the network layer protection, even if your server is locked away in a data center, the Internet connection can have a back door for hackers to access sensitive information and destroy your business. In this case, you need to adhere to industry best practices and tap some hardware authentication devices which can be more powerful than just software.
Continue reading our Cybersecurity Blog Series to learn more about how you can protect your application starting at the Hardware Level.
Edge and Cloud Computing
Business practices and processes today have probably forced many companies to implement some form of Cloud and Edge Computing. Internet of Robotic Things typically leverages these two to improve user experience, increase efficiency, and manage robot complexity.
As a result, you can probably install a server rack in the production floor or anywhere closer to the application to reduce latency and improve response times between the robot’s brain and body. This is particularly helpful in warehouses or mines that use Autonomous Mobile Robots (AMRs) or Automatic Guided Vehicle (AGVs). However, you need to take security precautions and mind harsh environment conditions.
As you can see, remote-brain robots are revolutionizing applications by allowing machines to execute complex actions like Deep Learning and Artificial Intelligence. These robots are only possible thanks to Internet of Robotic Things. Even though these robots pose new concerns due to their specific hardware requirements, if you partner with an experienced OEM, you can guarantee that your robotic application will run smoothly now and in the future.
If you want to continue reading about robotic applications, read our Robotics Blog Series.