Dan Knights Utilizes AI to Predict Emissions on Ships

Department of Computer Science & Engineering Associate Professor Dan Knights is utilizing artificial intelligence to predict the carbon emissions of ships on the ocean. Shipping accounts for a significant fraction of the total global carbon emissions, and most ships in the ocean are not currently required to report their emissions or fuel efficiency. Knights’ work is intended to help paint a fuller picture of the actual shipping emissions for individual vessels.

“It is a giant black box which makes it hard to do any attribution of emissions to different operators and owners,” said Knights. “Fortunately, there is a subset of ships that are required to report their emissions to the European Union. Of the 100,000+ ships on the ocean, we have data for about 18,000 of them over the course of several years. We have been using AI to train a model on this dataset in order to predict the emissions for the remaining ships.”

The goal of the project is to improve transparency in this sector as the world works to lower emissions in the fight against climate change. Knights’ group has worked with some groups in the nonprofit sector to work with this data, including Climate TRACE. Climate TRACE is a multinational coalition that was co-founded by former Vice President Al Gore, WattTime, and TransitionZero to enable real-time global tracking of carbon in every sector.

“Once we know who is emitting and how much, then that information can be used by regulatory bodies and consumers,” said Knights. “It is good to be aware of where improvements need to be made. Once you know which ships are the worst contributors, you can find similar ships to make improvements and recommendations. Other future work could include analysis of routes and putting together an optimal path and cadence for shipping.”

Knights’ project works with a number of interacting predictors that are nonlinear - enter the power of AI.  

“There is also a spatial element that could be part of this work going forward,” said Knights. “Ships are required to send out radio pings frequently, so you can actually figure out where ships are at any given time. Those data, combined with the miles per gallon data, weather conditions, etc. can help us estimate carbon emissions. All of this could eventually be done in real time.”

Much of Knights’ research has focused on using computational genomics to study the role of the microbiome in industrialization and human health. This work continues, alongside this more recent work focused on modeling sustainability of various practices in human civilization.

“All of my projects have an underlying theme of data mining and looking at complex datasets,” said Knights. “Generally, I am interested in looking at what society can do to improve our practices to make the world a safer and healthier place.”

Learn more about Knights’ work on his personal website