Artificial Intelligence, Visualisation and IoT for Environmental Science

Modern IoT (Internet of Things) data capture, storage and visualisation confront the developer and analysts with severe data challenges: massive volumes, rapidly changing, temporal, spatial, contaminated, wide, imbalanced (and so on). But the use of IoT environmental data has huge opportunities to help scientists understand the natural world, how the climate is changing, and influence policy makers [1-3]. This project will seek to develop novel and effective solutions where standard approaches are ineffective. The work will investigate theoretical and practical solutions that will transfer across from computing to application domains in computer science and environmental science and oceanography. One solution could focus on solutions to weakly supervise learning, within a full-stack framework. This could be used to extract actionable insights from the digital streams, reducing the volume of data stored, and enhancing visualisation. Consequently, the work requires expertise in coding but also full-stack development. We will work with sensors, and Bangor's network of LoRaWAN gateways linked to The Things Network, developing IoT Full Stack solutions that combine the use of Front End (e.g., HTML, CSS, JavaScript) with Back End development (e.g., Python, PHP) with Internet of Things (firmware systems, networking protocol, and sensors) underpinned with AI learning solutions. Experiments will be carried out to explore learnt analysis of techniques to highlight and store salient features and learnt behaviour and visualise the results to the end-user. We will work alongside various partners at Bangor University: electronics engineers to develop IoT sensors; and environmental scientists to enable the development of robust solutions for the challenges they face. This project, by developing an intelligent IoT pipeline, enhanced with AI, will provide the infrastructure to enable these challenges to be met.