Into the manufacturing matrix: IntellIoT and Industry 4.0
Those keeping an eye on developments in the manufacturing world will be well aware of the so-called fourth industrial revolution that awaits us. That is, a revolution in which smaller and more agile plants where products are created and assembled to meet exactly what the customer wants.
In these scenarios, cutting-edge technologies are harnessed that will enable a slew of machines to collaborate with one another to manufacture each product tailored to specific demands –doing away with the mass production-enabling factory lines in which machines churn out product after product in a system that does not allow for much uniqueness or originality.
It’s a technologically-complex and compelling upgrade from Henry Ford’s famed factory lines and their cookie cutter-churning offerings.
In these so-called “Smart Factories,” machines with an increasing degree of autonomy produce items as and when instructions are fed into their system by customers eager to own a product that exactly befits what they want.
This is the future of manufacturing for a variety of reasons: not least because its operations will be far more sustainable.
This is because, rather than churning out product after product in hope that there will be consumer demand, Smart Factories’ units begin production as and when an order comes in, an overhaul in how things are done that will cut down dramatically on waste and excess output.
The future is on its way, and it’s complex
While Industry 4.0 remains a futuristic vision– to some degree, the setup is already here, with some especially forward-thinking companies exploring how to harness such systems of human-to-machine interactions so as to allow the consumer to determine what they are getting. Berlin-based startup MyMuesli is already offering a service that allows the customer to opt for what they want from an assortment of products, without manual input of any kind. Another pioneer in this space is Adidas, which boasts a factory for highly customisable shoes.
But that’s just the start. We are still a long way away from being in a position to have increasingly autonomous machines in plants that are capable of collaborating with the agility and flexibility required to produce as per specific requirements. In order to make this vision a reality, strides need to be made at the cross-section of a breadth of related fields which IntellIoT is investing in.
This includes powering a web-based HyperMAS system (more commonly referred to as a multi-agent system)that can seamlessly orchestrate machine-to-machine collaboration, alongside the need for progress in the realm of edge computing such that systems enjoy the latency required to bring together the various agile parts of these processes of training collaborative and increasingly autonomous machines. What’s more, ensuring that the learnings gleaned via artificial intelligence are spread across plants is no mean feat, as engineers are put to the task of beefing up federated learning processes so that they can take place at scale. Throw AR and tactile relations technology into the mix, and one begins to recognise the extent of IntellIoT’s breadth and ambitions.
For Siemens’ Andreas Zirkler, who heads up IntellIoT’s Manufacturing Use Case, it is the act of bringing together all these diverging technologies that poses one of the greatest –and most interesting challenges of the project, as IntellIoT’s experts grapple with the work of coordinating and combining a slew of interrelated areas of expertise, exploring the possibilities that come with a cross pollination of ideas across the IoT ecosystem.
Arne Bröring, a senior researcher at Siemens in charge of IntellIoT’s technical management, agrees with Zirkler.
“Personally, I am excited that we are involving so many different technologies that are currently really hot topics of innovation,” says Bröring. “This ranges from distributed AI, to augmented reality for interaction between human and machines, to blockchain technology, which we aim to use in order to enable the integration of 3rd party service providers, for example in cases in which machines and robots in plants are actually owned by different providers”.
In bringing all these technologies together within the manufacturing context, IntellIoT’s researchers hope to help forment a host of new ideas and innovations, as pioneers work together to push the boundaries of the Industrial IoT space.
Humans and machines working together
IntellIoT’s bold ambitions in combining various cutting-edge technologies –and operating at the frontiers of all of them– are matched by its unique approach in exploring how to manage human-machine interaction across use cases. That is, working to create a seamless system in which machines appeal for assistance as and when needed.
The so-called ‘Human-in-the-loop’ component to the project operates on the principle that the AI in question is not foolproof and will regularly –most likely on a daily basis– face challenges that will need to be addressed by a human operator.
Examples of such events occurring include tasks like learning to have to pick up an item that seems unfamiliar, for example a vase with a strange shape– something that might confuse a machine in a Smart Factory, where assignments will require flexibility and adaptability in how its machines coordinate tasks in order to realise unique goals.
Here is where the human-in-the-loop comes in– when faced with an unforeseen challenge or task, the machine turns to a human operator for advice, which in turn teaches the machine what to do, with the help of Augmented Reality or Virtual Reality appliances that allow him or her to respond in real time at a spot remote from where the problem is taking place. Such feats won’t be easy to achieve, for a number of reasons, not least that the computation demands are high and will require groundbreaking work to be achieved in the realm of edge computing and federated learning.
The end result is a system so sophisticated that the human-in-the-loop can teach robots new tricks with simple demonstrations, serving as lessons that are distributed across increasingly autonomous agents operating through the plant.
Bröring gives an example of a scenario in which this human-machine dynamic will be important: “The robot arm needs to place a new kind of work piece in a machine and the AI is not yet trained for that. Then, IntellIoT aims to involve the operator of the plant and provide her with augmented reality glasses and a stylus to safely interact with the robot and teach it how to grab and place the workpiece,” he says. “After the completion of this interaction between human and robot, the AI can be re-trained and learn from this”.
Bröring is not alone in his eagerness to do his bit to help extend beyond the frontiers of the Industrial IoT space. Many enthusiastic engineers operating across the Consortium and across use cases share his passion. Zirkler also believes IntellIoT’s human-in-the-loop dimension will produce some of the most significant achievements for IntellIoT as for the broader IoT ecosystem across Europe and beyond.
Realising these goals of creating such a system requires work that will overcome all sorts of challenges, as tactile communications are leveled up alongside the computation power required to make such interactions happen at scale, and as engineers work the create ideal deployment of edge computers –goals that will keep our teams very busy through the three years of IntellIoT.
“With this human-in-the-loop concept, it’s really a big step ahead of what is possible today. So that is why it’s so exciting,” says Zirkler.
For more materials and updates regarding the IntellIoT project, subscribe to our newsletter.
For more materials and updates regarding the IntellIoT project, subscribe to our newsletter.
