Computer scientists enhance crowdsourcing techniques for sustainable disaster relief
Researchers from the University of Southampton are testing strategies that will sustain and improve crowdsourced analysis of drone footage in humanitarian relief efforts.
Professor Elena Simperl from the School of Electronics and Computer Science (ECS) is spearheading the new research project as it explores how to maintain crowd contributions after an immediate emergency has passed.
The experiment is a winner of the first wave of Collective Intelligence Grants from the Nesta innovation foundation, which seek to combine human and machine intelligence for social good.
Drones provide a cheap, fast and detailed means of capturing aerial images in disaster zones, especially when compared to helicopters and satellites. The analysis of this content can now be crowdsourced to people across the globe thanks to modern mobile technologies, enabling aid workers to coordinate recovery efforts effectively from the ground.
High levels of media coverage can attract a lot of volunteers to analyse drone footage during an emergency, but the number of participants can drastically decrease once the media attention fades, Elena explains. This reduced participation can have a troubling impact on communities long term recovery, with less reliable data making it difficult to keep information up to date and assign resources in an effective manner.
Southampton researchers have partnered with WeRobotics and Rescue Global for the project, which will draw conclusions that are relevant for all humanitarian relief, citizen science and crowdsourcing projects that struggle to maintain engagement.
Scientists will first explore strategies such as task variation and sequencing to test if they could sustain analysis in post-disaster and development scenarios. They will then investigate how experts and volunteers learn over time, acquiring skills and motivation to conduct more difficult analysis of drone footage.
Professor Gopal Ramchurn, of ECS, says: We have a strong ethos of working with practitioners and co-creating research questions and tackling these questions with them. WeRobotics and RescueGlobal will be helping us validate the crowdsourcing techniques we will develop within the project, bringing to bear their experience deploying such platforms in the real-world.
The Collective Intelligence Grants programme has received over 200 applications since its launch in September 2018, with researchers around the world proposing innovative approaches to solve social problems.
Twelve successful experiments have been announced in the programmes first wave, with recipients as far as Hong Kong and San Francisco launching experiments ranging from the use of swarm algorithms based on bees and fish for groups with conflicting political views, through to crowdsourced airstrike images as digital evidence for legal practitioners in court.