EXCLUSIVE: AI is powering evolution of the maritime industry. Lee Kok Leong, our special correspondent, finds out from Tarry Singh, a top artificial intelligent (AI) expert, how AI is driving the economic growth of the maritime industry to its next stage of evolution.
Tarry is CEO, founder and AI researcher of AI start-up deepkapha.ai, where he focuses on providing training to build AI departments, research and philanthropy. He is also an adjunct professor/guest lecturer at the University of Texas at Dallas, University of Chicago, Copenhagen University, Charité Berlin, and University of Catalunya, Barcelona.
“The latest trends and applications of AI or better said, Machine – and Deep Learning, are in practically all industries. We are seeing the theoretical concepts of Artificial Neural Networks (ANN) which was developed in the 1950’s, now becoming mainstream in almost all industry areas.
“The key word to remember is the “learning” part of machine or deep learning. We see today that learning to drive has led to a huge industry shift in automotive sectors such as self-driving cars.
“Next is learning to listen, translate text and reply back. This has revolutionized language translation, something which we consider magical when we see a robot answer in nearly complete sentences to the questions we pose.
“Also, another amazing learning example is the ability to further translate across languages, which many translation engines do “on-the-fly” and has a neural network in the back end doing the “intelligent” work. This field is continuously expanding. Just recently, Google has released a tool that can search for answers inside a large corpus of text. It is fascinating!
“This only gets even more exciting as AI is starting to enter traditional industries such as healthcare where algorithms learn to diagnose, predict and even prognosticate diseases such as cancer, among many other ailments. We see in manufacturing where AI goes beyond traditional analytics and more into predicting outcomes and scenarios.
“Select any industry and you will have AI dominating the news or media channel.”
Maritime Fairtrade (MFT): What are the impacts of AI on maritime industry?
Tarry Singh (TS): I spent a decade working in the maritime industry in the 90s in Japan, the US, Brazil and Europe. I even worked for a year with the Singaporean company NOL/NSSPL. I still remember writing my first software application in the mid 90s on real-time monitoring of vessel gas installations in visual basics.
After leaving that profession, I also participated in a leading Electronic Data Interchange (EDI) project for my first employer in the Netherlands where we developed software.
The maritime industry has always been slow in adopting new and emerging technology and I believe that AI may be an opportunity that shippers should be looking at very seriously.
Maritime operations have been extremely optimised but there is definitely that “last nautical mile” efficiencies such as vessel precision operations using various geographical data to exactly make ETAs/ETDs, fine-tuning container routing and re-routing, fuel-consumption models that offer “Fuel Savings Guarantees” like what Cargotec’s subsidiary Kalmar is doing.
Safety is a great area where accidents and casualties can be reduced using Machine Learning and Deep Learning techniques.
MFT: With regards to adopting AI, what is your opinion of the current state of maritime industry?
TS: We clearly see some winners such as Wartsila, AP Moller Maersk, Mitsui OSK and many other global players who have gone ahead and adopted at least the basic form of Machine Learning and further drive down costs while improve efficiency.
But I still believe that given the fact that AI is becoming increasingly available, there is tremendous scope for the maritime industry to make full use of this ready-to-serve technology.
MFT: What do you think are the major challenges for the industry?
TS: The maritime industry is an extremely low-margin business on its own. In 2008, the way to measure the global recession was to take a look at the BDI (Baltic Dry Index), which many economists also use as a “canary in the coal mine” for future signalling.
Having said that, there are major challenges that the maritime industry will be tasked to solve as the world population continues to rise in the coming decades.
Demand for food, energy and water supply will need transportation to be speedier, and use renewable energy and potentially novel ideas such as aquatic food.
The cost of transporting goods via ocean will need to continue to go down while the legislation will impose stringent limits to GHG (Green House Gases) emissions. This is bound to happen as global warming will continue to apply pressure on governments to cut down or wholly shift their air and road transportation costs drastically.
Another challenge is the traditional pace with which the maritime industry has adopted technology versus the pace of innovation that it will be hit by, such as by IoT (Internet of things), artificial intelligence, blockchain,, sophisticated sensors technology, advanced robotics and much more.
This all is commonly known as Industry 4.0, which will require the maritime industry to dramatically transform their workforce to learn or re-skill vis-à-vis new technologies. (Editor’s note: For a more in-depth discussion on this topic, see Tarry’s article on Forbes here.)
Another big challenge will be for executives to be more data-driven and focus on how and where they can drive data-driven sales and revenues using Machine Learning and Deep Learning technologies. (Editor’s note: Tarry had written an open letter to the CEO of Caterpillar regarding the importance of mapping a company’s corporate objectives with specific data-driven projects. Read it here.)
There are many areas of innovation but the maritime industry and individual shipping firms will have to make smart choices to adopt a compliant yet technologically advanced solution for a sustainable future.
MFT: So, what are your recommendations for the industry moving forward?
TS: The following are my recommended 6 steps.
- Pick areas of innovation starting with low-hanging fruits. Digitization is spurring automation and the future of a connected and self-managing maritime eco-system is getting even more real. Major benefits can be drawn from Machine Learning by simply identifying patterns.
These can be in the areas where there is a large-scale collection of data but not much is really happening with it. Smart automated activities such as selecting best alternative port or route, will help the shipping firms better prepare for their logistical needs.
- Use Deep learning to gain competitive edge where you can. It is an advanced form of Machine Learning that can be used in areas of computer vision, for instance for measuring crews’ physical and mental state.
Using simple computer vision can save a lot of lost hours, damages and in worst case, casualties. It can also be used during sea operations to continuously measure anomalies. In port or cargo operations, it can observe the behaviour of cranes, pumps and predict actions such as shift ballast tanks etc., automatically.
- Share your challenge with an open community. These days, you do not need to hire expensive consultants to solve complex tasks for you. You can just upload your challenge to a competition website such as Kaggle and you will see that thousands of developers will jump on your problem and solve it for very little cost.
For instance, Airbus launched a challenge a few months ago to “Find ships on satellite images as quickly as possible”. In this Ship Detection Challenge, participants were required to locate ships in images, and put an aligned bounding box segment around the ships they located.
- Setup your Machine Learning/AI lab. This is very important but which many organisations continue to ignore. Experimentation and building start with simple exercises as I mentioned above but to take this beyond just planning and into execution.
Shipping companies need to be more competitive and find difficult problems to solve. For that, it is important to have good awareness of where your data is and what can be done with it. This can only come with organised effort.
- Training your executives as well as IT staff about Machine Learning. Data scientists are the most sought after professionals in all industry domains. PhDs are in hot demand and there is a huge shortage of engineers in the industry. It is therefore very important that companies take a good look at existing staff and consider re-skilling those with the right attitude and background, as compare to hunting for talent from outside.
- Start with low-hanging fruits analytics projects. Go for both cost optimisations as well as growth. Many shipping firms have solely focused on cost optimization since the 90s but have never captured the opportunity to explore growth options for decades. This needs to change.
MFT: Your opinion on the future of employment, and relevant skills that we need to leverage on AI.
TS: There is a lot of uncertainty and worries that Machine Learning will replace jobs and this can be a cause of concern for many organizations. Such digital disruption happens from time to time. Employees need to focus on upskilling themselves in AI technologies.
As for the fears, I can assure you that neither jobs will disappear completely nor robots will become super intelligent. The most likely outcome will be a combination of “Human and AI Assistive Intelligence”. While independently, they may vary in the precision with which they do their tasks. However, the combined deep learning networks and human experts have a great future working together in a win-win relationship.