Towards autonomous shipping: Situational awareness needs sensors and communication

From a background in information technology, Negenborn got involved in the automation of shipping. “I am inspired to investigate if ship operations by computers can improve efficiency in logistics, energy efficient sailing and safety. If traffic management systems, like the operations of locks and bridges, can communicate with the systems onboard vessels, their speed can be adjusted to be just in time at a lock or bridge when it opens. This can ease handling of ships and also save fuel, because the ship is not speeding towards a transport hub where it would then have to wait. Improved safety could be realized by processing a lot of information that is acquired through many, many sensors by digital systems that with some impressive advantages: they don’t get tired, they will not get bored because of repetitive tasks and they can process much more information than a human operator can.”

From assistance to autonomy

While pilots are executed with remote ship operations from a shore-based station, now is the time to learn about the practical implementation of autonomous shipping. “Automation begins as a tool to help human operators make decisions with more information to base these decisions upon. Ships making their own decisions is a step that will follow later. The control software for vessels would not require the ship operator to specify precisely what the ship needs to do, but would let the operator provide more higher level goals, such as operate efficient, sustainable and safe while sailing between a certain origin and destination,” Negenborn explains. “These goals can go together and strengthen each other, but they can also cause conflict. Which of these goals will prevail when the system will have to choose?” The systems are self-learning, so their choices in specific situations are not pre-determined explicitly. Making the right choice will require the computer that controls all ship operations to think ahead, overseeing the outcomes of different choices along the duration of the journey. “This is an interesting dilemma,” Negenborn perceives. “In the process of transferring responsibilities from human to machine, can we trust the machine to make the right choices? Or will we learn that the machine is capable of making better choices?”

Combining data-driven and model-based approaches

From the perspective of software development, two approaches are possible for creating vessel control systems. A ship can be equipped with numerous sensors that register everything that is going on with engine control, rudder control and all of the ship’s systems, coupled to navigational information. The connected computer registers all of this data and sees what the human captain is doing, to learn which decisions are good and which decisions are bad. This is the data driven approach. Yet, in the maritime industry, ship models are available and a computer can be properly instructed about the particulars of vessels and the correct way to operate and navigate them using such models. You would call this the physical model-based approach. “Ideally, we could combine these two approaches when we create vessel control software,” Negenborn remarks.

Beyond navigation sensing the unexpected

“Situational awareness of a skipper is not limited to addressing issues with navigation. The quality of sensors that detect obstacles in the water, other vessels and more has improved. Yet, a skipper would also notice a strange scent coming from the engine room or an unusual rattle. Situational awareness includes being alerted by odd situations on board. To tackle such events, control systems need sensors to register it.”

“With the current state of automation aboard ships, there are still a lot of uncertainties. Computational capacity is sufficient to start investigating self-learning systems to take the decisions in case of calamities on board. Like humans, the digital systems need to be supplied with information on which to base decisions.” Sensors are essential. “If the vessel control system determines a malfunction in one of the systems that requires immediate repair before safe navigation can resume, the system is instructed to perform ‘graceful degradation’, to get the vessel safely to the nearest quay with the systems that are still running.”

Reducing crew dependency through automation

“Entrepreneurs in inland shipping indicate that a large part of the cost in their operations consists of wages for the crew,” Negenborn has learnt from sector inquiries. “In this aspect, some interesting projects are going on. For example, Roboat has developed an autonomous water taxi for the Amsterdam canals. Furthermore, a waste collecting barge was developed that can sail fully autonomous. The Dutch SMASH consortium, i.e. the Smart Shipping Platform, is working towards the realization of autonomous ferries and unmanned surface vessels.”

Within the framework of regulations for shipping, completely unmanned vessels may not be tomorrow’s reality. Digital support and remote sailing can however contribute to reduce the need for crew capacities as of today.

Learn more

Do you want to learn more about developments and projects in the Rotterdam region? Sign up and receive a number of valuable videos, white papers and articles.