Rise of the ethical Cobots at the Intelligent Edge (Part I)

by | Jan 4, 2021 | 6g

Credit : Mikko Mäkelä

Photo credit: Phillip Glickman.

In this blog post series we will sneak into the future and how it will be affected by the arrival of the 6G technology that is already in the making in Oulu, Finland, home of Kaira Clan. As opposed to the 5G technology, 6G will redefine networks infrastructure and roles of IoT devices, not simply by enabling much faster connectivity – again – but by changing the essential requirements as none of its predecessors did before. Join us in this blog posts series to discover the how’s and why’s but perhaps even more thrilling: the what if’s? 

Welcome to our second post on how 6G will redefine the experience of being a human on planet Earth.

Nope, I am not exaggerating.

If you read our first blog post you might still remember that feeling in your guts when the abysmal difference between 5G and 6G became undeniably obvious and pressing like: “OMG, we need a strategy for this for yesterday!”.

Today I am going to give you the chills. We are going to move from “pressing business need” to “existentialism” – and it’s only Monday!

But before that, let’s run a simple test. I am going to shoot some buzzwords that we will be all hearing a lot in the coming years and when I am done you tell me if your brain hurts or not.

Here we go: urban computing, liquid software, near memory computing, inference at the network edge, liquid computing handover, fog computing, generalized adversarial networks, knowledge distillation, transfer-learning, edge network digital twin, mobile edge, mobile fog, holistic orchestration, swarm intelligence, genetic algorithms, encapsulated individual microservices, federated learning, differential privacy, mean-field control theory, mutual platform verification, homomorphic encryption, mist computing, tactile internet, cobots and the almighty mother of all the mambo jambo techie headlines from the future: THE INTELLIGENT INTERNET OF INTELLIGENT THINGS. Fade to black. The audience goes bananas.

Yeah. You can take that paragraph above, put it inside a black bag, shake it and ask your children to pull out two random words and read them along. Repeat and you will get each time the fancy name of the Universe threatening technology that the USS Enterprise spaceship crew will try to defuse for each one of the following twenty Star Trek movie franchises.

C’mon. Admit it. If you are in tech this is fun. It is the butter and bread of our cryptic jargon based culture. We found it hilarious but this is just an indicator on how crazy and mind blowing the discussions around the 6G foundations and Use Cases are currently looking. These are not made up terms coined by tech journalists trying to write click bait headlines for geeks. These are the actual terms that the real scientists behind 6G are using on a daily basis.

If this is the ground zero vocabulary imagine how the future headlines of articles will read and how in a few years from now. We are about to evolve from “click bait” into “click acid”. By comparison, you have to admit that this article’s headline is damn conservative, not to say boring.  

But hey… how are you feeling? How is your brain going?

If it hurts then don’t be worried: you will enjoy this reading because we are not going to assume you know all those.

If it doesn’t then you are going to enjoy this reading too because we are going to connect all of them under a different light vs the typical paper, or the hyped mass media overselling the new tech (again). And, if you are geeky enough to be familiar with all those already by 2020, then I know you like new perspectives. Worst case, if those perspectives aren’t new to you, my personal goal is to make this post appealing to those people around you who do not share your excitement – yet – so you will recommend this reading to them.

Spoiler alert: they won’t read it but they will understand this is coming and they should start listening to you (they won’t, but hey, it is an “I told you” ticket!).

Let’s break that Intelligent Edge down, baby

So what is this all about? What is the Edge? I mean, you know about “Edge computing” (don’t you? Nope? Back to our first article, no rush, I will wait for you right here. You back? Ok, let’s go!).

Why does everything in tech have to become sooner or later “smart” or “intelligent”?

You might say.

We have been reading about intelligent buildings and smart appliances for ages but somehow all those promises feel empty. We are not living in a world where my home supplies magically walk into the closets and drawers saving us the first world problem of running out of toilet paper – and please nobody says “Amazon had a button for that“. This is not what we are talking about here. I still have to reboot my router every now and then USING MY HANDS (less often since I flashed it with OpenWRT firmware but still). If I lose my keys, or even worse, my smartphone, then I am done and Gods of Technology know Bluetooth headsets likelihood to spontaneously un-pair themselves is directly proportional to the importance of the incoming teleconference. I feel you.

Okay. I signed my last rent agreement digitally from the comfort of my sofa, and Finland is certainly more digital than many European countries – says DESI (the Digital Economy and Society Index by the European Commission); but this is not the Sci-Fi movie you, marketeers, have been feeding us for years, abusing the terms “smart” and “intelligent”.

So nope, I am with you dear reader: I am not buying this “x + intelligent” manipulative equation again. And this is my last word. Period. 

…

… but this time is the real deal – whispers the wind.

This is when the future finally catches up with your present. Yes. This is like when you suddenly realize: “how could we live without WhatsApp? Without smartphones in our pockets? Did the Middle-Age actually happen? Wasn’t there any YouTube before 2007? Gosh… what did they do ALL DAY???”.

Yes. That’s what we are talking about here. That is what the 6G and the Intelligent Edge are all about: a new breed of technology reshaping our daily lives so deeply that nobody will be able to understand how life was possible before it.

The first thing we need to be aware of is that the Edge is not a monolith, it is a fragmented entity that lives somewhere between the networks’ base stations and the smartphone in your hand. The Edge is the place between the Cloud datacenters – big and far away from you – and the TV on the bus station you look at every day looking for the next departures, or the credit card terminal accepting the payment for takeaway coffee. Those TVs, card readers, and other everyday objects used to be dumb, “processor-less”, but not anymore.

Now they are smart but isolated. They don’t talk too much to each other, and certainly they don’t have a proactive disposition. Today they are shy, overprotected by their host controller somewhere up in the network hierarchy, incapable of reaching their peers. If they could just talk to their friends, you wouldn’t believe what they could do for you… suddenly these inanimate objects will come into life around you, quite often invisibly contributing your daily user experience in ways that will make you feel like you’re still dreaming.

Those smart but confined devices have been replacing their dumb ancestors and nowadays populate the space between the Cloud and you. A space that can be divided into fog and mist as we get closer and closer to the final user.

The user is part of the mist and sometimes might be part of the fog, the lines are blurry here – as you might expect from a real mist and fog. The frontier between mist, fog and cloud depends on how participant our reference device is in the process.

Let’s take a look at what we can consider dormant components of the future mist, e.g.: the microcontrollers inside cables, power adapters or touch panels.

Silently and deviously even our power adaptors and cables have been incorporating some degree of intelligence. Many cables are no longer “just a cable” since the information they carry is overseen by built-in microcontrollers. These microcontrollers are  – by today’s standards – simple computers manned by equally simple processors, which are actually more powerful than any desktop computer from the early 80s.

A painstaking example of this are hackers porting vintage 3D games code to run on the Touchbar of Apple laptops. Those games are effectively running on the obviously underutilized microcontroller responsible for handling nothing more than our greasy finger interactions; but they can do more, much more! And imagine how much they could do if they could coordinate with other underutilized fellows around them. But they can’t. They can’t do that because the standard for that conversation has not been agreed upon by the industry, yet…   

In a nutshell: your smart cable can’t talk to your laptop’s touch bar and none of them can ever meet that sexy smart toaster in your kitchen. They are all smart appliances with a real and very capable processor inside but nobody is harvesting their power let’s not even dream about orchestrating and coordinating their efforts into some sort of swarm intelligence.

Why would anybody do that? You could ask. Simple. Because through your daily interactions these devices know you very well. Actually and some cases even better than your family and friends. They know your routines, and their records of your apparently mundane interactions are the holy datasets that can make the Intelligent Edge – of which they are all part of – to even feel your mood, e.g.: behavioural inference from gesture patterns.

5G already had an Edge! How is it different from 6G’s?

5G’s Edge was about putting computing power at the Edge of the network, or harvesting the existing one. That is making sure your existing or new connected devices around you (routers, appliances, peripherals, …) have access to a fast enough network.

There is a lot of data to be crunched around 5G Edge devices, ideally not all that data must be shipped to the almighty Cloud, even with the breathtaking 5G speeds if we start broadcasting all the data from the Edge, raw and unfiltered, there is a good chance to collapse the network and never getting a proper response from the equally overloaded Cloud.

And even if the 5G network would be capable of handling that workload it is not the most efficient way to proceed. Better and more timely responses can be provided if we infuse some degree of cooperative optimization among those devices. For many Use Cases, specially those requiring local sensors and metering, the quality of the dataset can be vastly improved when it is enriched by adding the measurements from other devices nearby. This requires a certain level of preliminary AI processing before asking the distant Cloud AI for its opinion; eventually the Intelligent Edge will learn to provide correct, accurate and reliable answers without even waiting for the Cloud. For many applications this level of responsiveness draws the line between a satisfactory and an unsatisfactory User eXperience.

If properly balanced, the Edge plays a key role to distribute, assign and escalate this operations across mist, fog and clouds.

We can make an analogy with any organization’s chain of command or even our own bodies: some everyday decisions are made by the nearby cells without involving upper systems. We do not consult with our upper brains (the Cloud) every blink of our eyes, every breath we take, every step we make, I’ll be watching you. Yes, Police. Just checking you are still reading this. But you get the point. 

5G does not say a word on how this can be achieved. The 5G standard does not regulate how those devices are supposed to share and balance the workload of finding patterns among constantly changing streams of data. A dynamic exercise that involves an AI balancing AI processes.

5G and Edge computing were all about muscle power.

6G and Edge Intelligence (EI) are all about putting some brains behind those muscles.

5G is first and foremost a communication standard, not just a fancy new faster better piece of technology, it defines how the network is handled, and the elephant in the 5G room is that it does not include any specification, definition or requirement on how the network operators and users are supposed to interact, and benefit from, things like: Artificial Intelligence, deep neural networks, machine learning. The 5G network standard does not say a word about them.

Sure. You have read about 5G and AI but always in the context of what 5G could do for AI, and not the other way around.

What an Intelligent Edge can do for the 6G network?

The 5G ship has already gone. 5G networks are already being deployed. The 5G standard’s status has been frozen and blessed for public release and adoption all around the world.

And it’s good, you can’t hold these standards forever. 4G was already showing symptoms of overload and it was not nearly enough to satisfy the requirements from the emerging and very needed Use Cases. Which otherwise saw all the other required technologies mature enough to benefit from a new iteration of the standard: AR/VR for mission critical telepresence scenarios (remote surgery), drones, self driving cars… the usual suspects. The problem is that all those Use Cases have more complex requirements than just “faster than 4G please”. Therefore 5G is good, but it won’t be good enough forever.

Adding the whole AI’s toolbox to the 6G cocktail 6G provides the necessary understanding for all the hardware and software developers out there to intertwine all their connected gizmos, sensors, devices and materials into a new self-conscious and self-balanced fabric. A new breed of network that will autonomously cooperate to provide timely and accurate enough data improving the user experience to such a point that there won’t be anymore “user experiences” but a seamlessly, continuously, AI-assisted life.

The ultimate goal of technology is to disappear. To become so advanced that it feels like magic – paraphrasing Arthur C. Clarke’s quote. It will just be.

What the Intelligent Edge will do for the 6G network? Simple answer: it will make the concept of network disappear. You, us, them, all the things and the whole planet will become THE network.

What is a cobot and why are they supposed to be ethical?

A cobot is a relatively new term for “cooperative robot”. The typical robotic arm at the typical car manufacturer used to be conceived for working at fenced environments where the human operators should never enter unless they want to be welded together with the car. A cobot is supposed to be able to cooperate with humans in a safe way. Your Roomba is a good example of a cobot.

If we remove the requirement for having a physical body then any Intelligent Edge device could be considered a cobot since they also cooperate with us, virtually.

Under that light: all those smart cables, touch bars, smart gateways, routers, printers, appliances, the entire Internet of Things and its blue collar relative – the Industry Internet of Things – are, as of today, an army of cobots awaiting orders.

Those cobots on the edge, and their beehive AI mind, will interact with us via our data and they will decide, consensuated with their neighbor cobots, on what’s best for us.

To make this edgy data workload juggling even more complicated: we have been playing around with AI long enough to understand the risks associated with biased AI.

AI tries to emulate the human brain capability to find patterns out of unstructured data. To achieve that there are a number of tools we do not need to describe right now but you have heard about (Machine Learning, Neural Networks, algorithms…). The important bit is that digital entities learn through training. Training involves exposing them to datasets that are mere reflections of our imperfect real world.

If we agree that there is racism, sexism, injustice, and other maleficent behaviour in our world then we have to expect those to be reflected in our observations.

AI scientists have validated through due research what it might sound like intuitive evidence to anyone: you are what you eat. And AIs are no different.

Will this cobots’ constant learning and real-time decision making be fair, transparent and accountable? Or will they subtly lean our ways towards a maleficent unethical behaviour without even noticing it?

We will talk more about these questions in the second part of this blog post.

Stay tuned if you want to know how to raise a good cobot that knows how to behave without messing up its human overlords’ businesses.

This blog post a is a digested and Sci-Fi spiced up version of the +30 pages long great paper “6G White Paper on Edge Intelligence”: Peltonen, E., Bennis, M., Capobianco, M., Debbah, M., Ding, A., Gil-Castiñeira, F., Jurmu, M., Karvonen, T., Kelanti, M., Kliks, A., Leppänen, T., Lovén, L., Mikkonen, T., Rao, A., Samarakoon, S., Seppänen, K., Sroka, P., Tarkoma, S., & Yang, T. (2020). 6G White Paper on Edge Intelligence [White paper]. (6G Research Visions, No. 8). University of Oulu. http://urn.fi/ urn:isbn:9789526226774 6G Flagship, University of Oulu, Finland June 2020