Humans have been exploring the ocean for thousands of years, but now the power of AI can help unlock its mysteries more than ever. To commemorate the 400th anniversary of the Mayflower’s trans-Atlantic voyage, the Mayflower Autonomous Ship (MAS) will repeat the same journey—this time without any people onboard. The world’s first full-size autonomous ship will study uncharted regions of the ocean, and an AI Captain will be at the helm.
From Plymouth, England to Plymouth, Massachusetts, the crewless vessel will use explainable AI models to make accurate navigation decisions. The ship will collect live ocean data, delivering valuable research that can inform policies for climate change and marine conservation. Through IBM technologies, MAS makes all of this possible by advancing three areas vital to a successful mission: talent, trust, and data.
The future of the ocean is at stake
More than 3.5 billion people depend on the ocean as a primary source of food, and ocean-related travel makes up 90% of global trade. Since the 1980s, however, the ocean has absorbed 90% of the excess heat from global warming, endangering life both below and above the seas.
Protecting the ocean starts with understanding more data about its ecosystem, but this undertaking requires massive investment. MAS reduces the need for enormous resources in ocean research by using data and AI to augment human work (talent), navigate safely while meeting maritime regulations (trust), and fostering collaboration to develop actionable insights (data).
Talent: Saving time and costs for scientists
A typical ocean research expedition can take six weeks with as many as 100 scientists onboard. Only one week is often spent on actual research. The rest of the time entails traveling to and back from destinations and sometimes managing bad weather and rough seas.
“Traditional research missions can be very expensive, limited in where they can explore and take a long time to collect data,” says IBM researcher Rosie Lickorish, who spent time on RSS James Cook as part of her Master’s in oceanography.
MAS significantly cuts down time and costs for scientists. A solar-powered vessel, it travels independently to collect data in remote and dangerous regions of the ocean. Researchers back on land can download live data and images synced to the cloud, such as whale songs or ocean chemistry detected by an “electronic tongue” called HyperTaste.
“With AI-powered sensors onboard that can analyze data as it’s collected, scientists can access more meaningful insights at greater speed,” says Lickorish. “The cost of data for our experts is low, in time as well as money.”
Trust: Navigating accurately with explainable AI
A combination of technologies helps MAS travel with precision: a vision and radar system scans the ocean and delivers data at the edge; an operational decision manager (ODM) enforces collision regulations; a decision optimization engine recommends next best actions; and a “watch dog” system detects and fixes problems.
This entire system makes the AI Captain intelligent, allowing it to make trusted navigational decisions driven by explainable AI. Rules-based decision logics in ODM validate and correct the AI Captain’s actions. A log tracks exactly which initial conditions were fed into ODM, which path it took through the decision forest, and which outcome was reached. This makes debugging and analyzing the AI Captain’s behaviors vastly easier than the “black box” AI systems that are common today.
Safety and compliance are key. For example, decision optimization through CPLEX on
IBM Cloud Pak for Data, a unified data and AI platform, helps the ship decide what to do next. CPLEX considers constraints such as obstacles; their size, speed, and direction; weather; and how much power is left in batteries. It then suggests routes to ODM, which validates them or advises another course.
“ODM keeps the AI Captain honest and obeying the ‘rules of the road,’” says Andy Stanford-Clark, IBM Distinguished Engineer and IBM Technical Lead for MAS.
Data: Fostering collaboration for better insights
Once the mission is complete, researchers will use IBM Cloud Pak for Data to store data, apply governance rules to enhance data quality, manage user access and analyze data for actionable insights.
Having all data managed by a unified platform can enable greater collaboration for various project teams across ten countries. In addition, organizations and universities around the world can partner with the research teams, forming a grassroots coalition to advance measures that curb pollution and climate change.
Source: ibm.com
Posts
0 comments:
Post a Comment