The maritime industry - taking the human out of the loop
Posted on 15/11/2019
Artificial Intelligence and data is one of the pillars of the Industrial Strategy Challenge Fund (ISCF) and has included funding to advance ‘robots for a safer world’ and ‘self-driving cars’; both intended to take the ‘human out of the loop’, in the first case out of potentially harmful or harsh situations and environments, and the second in order to reduce or remove human error. Both are aimed at improving safety.
At the March 2019 workshop exploring the challenges from the Maritime 2050 strategy, autonomy and digitisation emerged as a central and cross cutting theme; able to contribute to solving issues raised around the environment and emissions. Autonomy was also the subject of the technology and innovation route map that accompanied the strategy in January 2019. It set out the opportunities as follows:
“Advances in technologies such as sensors, data analytics, and machine learning, mean that in the future, vessels could operate with fewer crew on board. Robotics or algorithms could perform tasks to augment human capabilities. Alternatively, vessels could be completely unmanned, and instead ‘crewed’ from shore-based control centres. Another possible development is that vessels could operate independently of any human operator, with decisions made entirely by machine.”
Projects examining these scenarios are taking place across transport and mobility, and also in areas unrelated to the ‘vehicle and driver’ element, such as those scoped in the recent Network Rail SBRI challenges on security surveillance and tunnel inspections requiring AI-based solutions.
A recent KTN internal workshop took place to gain a better understanding of the meaning and use of autonomy, together with the challenges involved and opportunities for cross sector working.
The benefits of autonomy and digitisation were reviewed – improved safety, reduced costs, increased productivity, standardisation and greater consistency, optimisation, greater efficiency, increased mobility, etc. It was also noted that autonomy could extend and enhance human capabilities, assist in decision making, and offer new functionality and capability, by joining together complex data sets and systems.
However, questions were raised as to whether decisions made by AI systems, and the basis for those decisions,
- could be understood by humans;
- take into account an understanding of all factors;
- and should be made on all or particularly ‘sensitive’ issues.
Three of the challenge areas discussed were:
1. Data. Too much data compounded by a lack of resource, and resource with the skills and understanding, to know what is important or unimportant. In addition, an unwillingness to share data is said to inhibit gaining an holistic view of an issue, or that simulated data when used to develop the algorithms, may result in ‘false positives’.
2. System interoperability. A lack of interoperability between the multiplicity of systems on board transport vehicles, or in use within and between organisations, is seen as a barrier to effective connectivity, the introduction of automation and any resulting potential productivity and efficiency gains.
3. Cyber security. The growth in volume of data, the sharing of that data between various parties and across the various fixed and mobile platforms, the hosting of those systems and data by third parties, the connectivity of the devices themselves, and with digital and automated infrastructure, present a challenge in both maintaining an overview of these complex systems and safeguarding them from ‘attack’.
You may also be interested in the current MarRI UK Technology & Innovation in Maritime call. This has a significant autonomy component, and is open until 20 December for Expressions of Interest.
The MarRINav project results workshop on 5th December in London may also be of interest: the MarRINav project aims to explore the vulnerabilities of Global Navigation Satellite Systems (GNSS) as a Position, Navigation and Timing (PNT) solution in the modern maritime environment.