Announced in April, the collaboration is expected to nurture over 200 AI developers by the middle of 2019 and focus on the three key areas of AI, the Internet of Things (IoT) and blockchain, according to OOCL cio Steve Siu.
Read more: OOCL teams up with Microsoft on AI
Setting the backdrop for the kind of change that is needed Siu told Seatrade Maritime News: “Shipping’s business processes evolved from the telex age where you operated on a very limited set of information because you could not transmit so much information.”
He went on to add that with better monitoring and transmission technology, and exponentially more data processing power, many more data points can now be used. This is the basis for the kind of innovation that OOCL is aiming to bring to the industry.
Introducing himself as the person who, at the turn of the millennium, brought fundamental change to the way OOCL used IT to run its business, Siu said much more dramatic change is in the offing.
Dismissing the suggestion of small technological improvements such as Robotic Process Automation (RPA) or other means to gain incremental improvements in efficiency, Siu elaborated that OOCL’s area of research cooperation with MSRA is in deep learning to help boost network optimisation.
Pointing out that OOCL was the first Hong Kong company to be invited to visit Microsoft’s research centre in Beijing, Siu said: “The area we are working on is fresh; basically it’s research, we are using the deep learning method to try and qualitize this area.”
While there are some clear targets, the end result of the project is as amorphous as it is critical to the future of the industry. Frankly acknowledging that top line Maersk is doing something very similar with its Maersk Digital platform, Siu said the jury was out on who would achieve a better outcome sooner. Maersk has tied up with IBM to develop an open platform for sharing information and developing digital products related to trade.
Microsoft Hong Kong gm Cally Chan noted: “We can all agree that digital transformation is an ongoing journey for many. In a digital-first economy, business challenges will become increasingly complex. It fully applies to maritime and shipping industries, where operations depend a lot on multiple variable factors – from weather conditions, congested ports, equipment breakdowns to fluctuating trade volumes.”
She added: “Fundamentally, we see how emerging and advanced technologies such as AI and IoT could address those challenges.” For example IoT could be used to continuously collect data on all voyage-impacting factors, both onboard and off the vessel. Meanwhile, predictive analytics through machine learning (ML) and artificial intelligence (AI) would lead to operational adjustments and decisions to keep vessels moving in the most cost-effective ways.
These insights could help masters cut fuel consumption by optimising the vessel’s speed while course or time to destination could be amended to take into account adverse factors such as anticipated port congestion or deteriorating weather, she said. Other applications include increasing operational flexibility and coordination across multiple vessels, improving transhipments, avoiding logjams and suggesting ways of revising routes to increase profitability.
Optimising shipping operations
“We will help OOCL optimise its existing shipping operations. Also, our partnership with OOCL will help us better understand the needs of the entire sector and provide the services our customers need the most,” Chan added.
This is echoed by OOCL’s Siu who pointed out that what the line brings to the table is its specialised domain knowledge. “They picked us because we have experience in this data set, so it’s not just raw data and bottom up learning; It is organised learning with domain knowledge,” he said.
“There are many tasks in ocean transportation (and in the logistics industry in general) that can be greatly enhanced by AI technologies, such as demand forecasting, resource positioning, demand dispatching, dynamic pricing, and service network design. The current solutions to these problems heavily depend on heuristics or domain-specific feature engineering,” said Microsoft’s Chan.
She added: “The collaboration with OOCL will allow us to use deep learning, reinforcement learning, and other advanced AI technologies to provide new, end-to-end, data-driven solutions to solve these challenges and optimize its operations.”
Cutting-edge AI solutions
She pointed out that MSRA has been conducting research on AI since 1991. MSRA will support OOCL by providing cutting-edge AI solutions to drive digital transformation through advanced AI technologies as well as help the line in developing next generation AI talents, where some 200 AI engineers are expected to be trained in the coming 12 months.
“We are extremely excited with the unique opportunity to jointly train 200 AI engineers within OOCL, whereby MSRA will take the lead in conducting machine learning and deep learning sessions at the Hong Kong Science Park over the next 12 months,” said Chan adding that the partnership is part of Microsoft’s commitment to training the workforce of the future.
“Deep and reinforcement learning is still an R&D area. There is still much to be learned by our top AI researchers from MSRA. In the meantime, OOCL will provide invaluable domain experts for business impact validation of these cutting edge deep and reinforcement learning advancements,” concluded Chan.
Finally suggesting that part of the goal might be to develop the network control centre incorporating all the current and future technologies, which he had envisaged years ago but could not achieve due to limitations in the technology then, Siu was philosophical however about the longer term prospects for the project, given that Microsoft has earlier this year announced a sweeping reorganisation while OOCL’s acquisition by China Cosco has just been completed.