Virtual Personal Assistance at Sea
Shipping has struggled to get to grips with data, but the tide is turning. The advent of reliable, high-speed, high-throughput satellite connectivity at sea means transferring the vast quantities of data is no longer the problem it once was. The challenge today is sifting that data for insights that can make vessels safer and more fuel efficient.
Ship owners and managers are also contending with a rapidly evolving regulatory landscape – particularly in terms of environmental performance – and vetting requirements from charters and cargo owners that are ever more demanding.
Fortunately, the advances seen in artificial intelligence, machine learning and related analytic techniques are now as available to shipping as any other industry.
At METIS, these technologies have been deployed to create a platform for capturing, processing and analysing data, with particular attention paid to making the outcomes of most relevance to and easily accessible by those using them. The user benefits from shipping’s first Virtual Personal Assistant, an ‘agent’ powered by Artificial Intelligence which – like a seafaring Siri or Alexa – can help seafarers, shore-side technical colleagues and executives alike to perform day-to-day tasks.
Creating such a powerful solution relies on addressing a major obstacle to facing digitalisation in shipping that does not block the route of other industries: lack of standardisation, in terms of equipment specifications and configurations on board.
Stepping on board any modern vessel offers a chance to find equipment from innumerable manufacturers, most of which opt for proprietary interfaces. The situation contrasts starkly with aviation, where airlines largely choose between just two manufacturers. Shipping’s apparent abundance of choice has, in reality, hampered integration efforts.
Finding a way to deliver this data that can be pooled, accessed and manipulated for multiple purposes as ‘big data’ has become something of a ‘Holy Grail’ for maritime digitalisation.
The METIS solution is to deploy a collection of smart devices, called Wireless Intelligent Collectors (WICs), which are meshed in a robust wireless network. The data collected is transmitted back to a cloud-based hub, where it is stored securely alongside information from other sources, like weather providers and traffic monitoring services including AIS, as well as corporate resource and maintenance planning. Everything is immediately available to the micro-services and ‘agents’ providing functionality and interacting with end-users.
The ability to gather data around the clock provides a much richer, high-resolution picture of vessel performance than would be possible through trying to join the dots using incomplete, patchy datasets obtained from noon reports or manual readings.
WICs don’t just function as a dumb relay, forwarding data unchanged as it arrives. They contain embedded processing power to carry out local pre-processing, cleansing and performing preliminary analysis of incoming data before it is transmitted back to the hub. By intelligently removing unimportant ‘background noise’, the architecture also reduces the load on the hub and improves overall system resilience and reliability.
This supports ‘data-driven decision-making’, helping staff to generate various operational scenarios quickly and easily. Based on the current condition of the vessel and dynamic parameters such as speed, ETA, draft, etc., METIS can help shipping companies choose the optimal course of action, without captains/superintendents having to resort to guesswork.
This qualitative approach contrasts markedly from the one-size-fits-all model previously seen in the maritime space, where unfiltered data has left manipulation or interpretation up to the individual user.
Recently we had a case of a real vessel faced with heavy weather on a passage from Malaga, Spain to Rotterdam, the Netherlands. The normal course of action is to slow the vessel down and let the storm pass before proceeding; running multiple scenarios on METIS indicated – counterintuitive – that less fuel would be burned by outrunning the weather.
Against this background, agents can be created to perform specific tasks: weather analysis, monitor hull fouling, measure fuel oil consumption, quantify onboard energy usage, track main engine operating efficiency. Agents can also adapt intelligently to their immediate environment, selecting the best available data source for the job at hand – whether from equipment sensors, manual readings or external information.
Because no two ships are identical, agents are also highly configurable and can be set up to detect and act on events based on vessel-specific thresholds. They have both diagnostic capabilities that can help engineers pin down the likely cause of an anomaly, and prognostic capabilities for predicting future outcomes so that corrective actions can be planned in advance.
Thanks to their predictive capabilities, agents can forewarn crew of potential impacts on vessel operation caused by changes in equipment status or forecast for the external environment. This helps to schedule maintenance tasks and aid voyage planning, also monitoring and evaluating machinery performance for optimisation purposes. The ability to intervene before an issue escalates offers benefits for safety, as well as the bottom line.
In exceptional cases where more rigorous investigative work is necessary to diagnose the root cause of a problem or make a decision, METIS can prepare and provide data in a format suitable for tools like Tableau, Power BI and MATLAB.
The way in which agents convey information to vessel crew and shore-based staff also represents a sharp break with industry norms. Instead of requiring another standalone application that complicates existing workflows, the default output channel for METIS is through the shipping company’s existing collaboration platform – think Microsoft Teams, Skype or Slack.
If necessary, the Virtual Personal Assistant can reach personnel more directly by sending text messages to alert them to critical events. Such updates and alerts are presented in Plain English as opposed to indecipherable error codes. Crucially, METIS won’t spam everyone in the organisation about, say, a broken fuel pump. It knows who needs to know what, and, depending on the event’s seriousness, when they need to know.
What is more, communication with the agent is interactive: recipients can respond, requesting more information or selecting a course of action – again in natural English, in much the same way they would chat with a human colleague. In short, anyone can now use data in their work without having to undergo training on how to code or query a database. This should enhance collaboration between different teams within a shipping company.
In this way, agents also assume the role of an extra pair of eyes for personnel in the engine room or PA for staff tasked with preparing regular performance reports. Against a backdrop of shrinking crews and growing administrative burden on those who are left, the value of such additional ‘helpers’ quietly and tirelessly supporting seafarers cannot be understated.
Source: By Mike Konstantinidis, CEO of METIS Cybertechnology