Siri, design a ship for me, please!

I’ve been wanting to start the Machine Learning course on Coursera for quite sometime, but it’ll take me a while before I finish off other things to move to it. I’ll share my half-baked views nonetheless.

So I was asking my phone the other day - “Siri, show me stock prices for Apple!”

And voila! Siri pops up with what I asked of her! She seems to be hell of a woman who is turning better by the day, rather than becoming older and forgetful.

Isn’t that the premise of machine learning? The more the data that we feed into the system, the better it gets at identification and recognition. 

Very recently Mark Zuckerberg completed building an AI assistant called Jarvis as his personal project (https://www.facebook.com/notes/mark-zuckerberg/building-jarvis/10154361492931634/). It is not surprising that the best companies are investing into self-learning systems including - but not limiting to - speech recognition, language processing and deep learning.

Until now, IT departments I’ve known at shipyards pivot around installations of proprietary softwares; server setup and maintenance, network security; and, once in a while, changes to systematic workflows through digitised changes. 

Elon Musk in his biography is not wrong to point out, that the best brains are pursuing finance, internet and law. But what if several of these folks got enticed and entrusted with building the next generation of shipbuilding systems.

“I think there are probably too many smart people pursuing Internet stuff, finance, and law,” Musk said on the way. “That is part of the reason why we haven’t seen as much innovation.” 

With a ship design, at any given moment, there are as many as 10 teams working cross-functionally on the design phase. These may be teams for marine systems, electrical and automation, hull and superstructure design, quality assurance, weapon systems and defense capabilities. Firms such as ShipConstructor and Aveva Marine have been trying to introduce innovations and facilitate seamless workflows towards a complete ship design, but as a whole, the industry is not moving fast enough as the advancements in counterpart industries of automotive and Internet.

Let’s look at the number of patents over the years.

Do we see that boom around 2000? Yes, that was right before the dot-com bust, where everybody wanted a piece of pie of the billions of dollars floating in the market with their new inventive internet technologies.

It will not be wrong to say – the next generation of systems – aided by computer technologies – are going to cloud computing, 3D visualisation, machine learning based systems and internet of things for interconnected systems (if leaving aside additive manufacturing which doesn’t really rely upon computing so much). 

So, what are the possibilities that we can have our own customised assistant to help us design and build ships?

An assistant you can ask questions or expect actions such as:

• What are the changes to the ship block B9 since 1 January 2016?
• Find clashes of electrical and piping systems in the engine room.
• Route the sea water pipe in the Pump Room from start to end point.
• How many brass-bronze gate valves are used in this ship?

Rant, imagination, wannabe wishlists? Yes, Yes, Yes. But the point right now is not discussing whether this is wishful thinking or hoping that these technologies will walk to us, but rather, how to accelerate us getting there?

Making shipyards digital is one way, creating core structure of all our systems to be adaptable to future learning algorithms and available for identification and recognition is another.

The first one can be achieved by, of course, digitalising our work processes from forms to processes and thereby creating easily accessible data repositories, and investing heavily into ERP systems.

The second needs much more than that. Since it is about rethinking design in terms of items and making it accessible for future systems.

In XML, Javascript, there is a concept called the DOM method.

To better visualise this, imagine me. Well, I’m a person with a multitude of characteristics, each one being an item. 

Let's say that you ask me what is my mood today? The system is able to look into the hierarchical tree and look up my mood element and return back its active value. Ujjawal is "curious" today! Well, hell yes, he is certainly doing a lot of abstraction today.

Ship design is also very hierarchical in nature, with a lot of inter-dependencies. We start off at the basic design and keep adding extensions of components.

Now, with each component defined, and ability to access them by integrating speech recognition at their hierarchical level, can we not recall them easily? Also, by defining these tasks every time and having the computer learn itself - these abstractions would become realities.

For example - I tell the computer to draw an isometric drawing and then show it how to do it. By ascertaining the keywords, the system is able to do it on its own next time. This is part of the training and tasking method for machine learning.

Going back to the same questions as posed before:     

• What are the changes to the ship block B9 since 1 January 2016

Check: Revisions, In: Block B9, Since: 1 January 2016

• Find clashes of electrical and piping systems in the engine room

Action: Clash Check, Elements : Electrical System, Piping System, Location: Engine Room

• Route the sea water pipe in the Pump Room from start to end point

Action: Automatic routing, Element: Sea water pipe, Location: Pump Room, Coordinates : Start to End Point

• How many brass-bronze gate valves are used in this ship

Element: Gate valves Type: Brass bronze Check: Number

Right there! We have our own version of Siri helping us complete day to day tasks and automate processes while we design ships. This however, is supervised learning, and requires another layer of accepting and rejecting actions and training the system to identify the correct elements and perform requisite checks and balances – which is a talk for another day.

It’s just about time to lure Silicon Valley engineers into the hallowed offices of shipyards if we want to step up our game.

Game On!