Opinion: Advances in AI and satellites can help us meet the SDGs

Imagine a world where breakthrough technologies and the most advanced innovations are used to keep people safe, protect forests, and help communities in crisis. Imagine governments, businesses, nongovernmental organizations, donors, and communities empowered, not only with reliable, regularly updated data for populous and remote areas, but also with tools to find a signal amid noise, draw attention to decision points, and enable efficient, effective action. We are almost there.

New methods that employ machine learning and artificial intelligence to make effective use of the incredible volume of available data from satellite images and the growing universe of other available sensors enable us to make giant strides toward achieving “peace and prosperity for people and the planet,” as laid out in the Sustainable Development Goals.

Alleviating poverty and ensuring a healthy future for all requires comprehensive understanding of populations and resources — and their complex interactions — and a way to track the impact of interventions. We’ve made significant progress in recent years in laying the groundwork. High-resolution, frequently updated images of Earth taken from space, together with rapidly advancing analytical possibilities using cloud computing and AI, can help prevent disease outbreaks, enable supply chain transparency, support efficient post-disaster response, map populations, provide insight into gender inequality, and disrupt human trafficking networks. The possible applications are as numerous as the interactions between the SDGs — that is to say, countless.

Preventing an outbreak

In May 2018, Ebola cases were reported in Equateur province, Congo. Days later, a new case in the province was reported from Mbandaka — a major transport hub along the Congo River, 150 kilometers from the original cases — threatening the spread of the virus internationally. Almost all existing maps of the outbreak zones contained inaccuracies, with some towns having never been mapped. How do you stop an outbreak without knowing where the people are or how to get to them? This was a major obstacle for containment efforts and health organizations working from disparate maps and sources of information.

Using a combination of machine learning-based algorithms and very high-resolution satellite imagery, Maxar — a global provider of advanced space-based technology solutions — created incredibly accurate building footprint maps of the 130,000 square kilometers of Equateur province. While the existing maps — which were frequently patchy and outdated — had required significant labor to create, the building footprint maps were created in just days. With accurate maps, in-country partners like PATH and Doctors Without Borders could work with the Ministry of Health to ensure data was put to immediate use to contain the spread of the disease.

Maxar developed this technical approach specifically for global health interventions with funding from the Bill & Melinda Gates Foundation. From crowdsourcing to the introduction of machine learning techniques, the approach has been tailored over several years for public health use cases in low- and middle-income countries. The approach is highly accurate, and it accounts for the incredible diversities across unmapped communities so that different structures — from thatched-roof huts to high-rises — are detected by the algorithm. For any humanitarian response, it is critical that technologies do not inappropriately or unintentionally bias the distribution of resources and services. Instead, they can offer an objective lens onto remote and often vulnerable communities where interventions can make a difference. We currently have an unparalleled ability to create rapid and timely data in response to a crisis, allowing aid agencies and governments to focus on the mission at hand rather than on filling information gaps.

Transparency in supply chains

Through its forest conservation policy announced in 2013, Asia Pulp & Paper committed to ensuring that no natural forest would be cleared in the production of its products. It’s not an easy promise to keep, given that APP sources from a number of pulpwood suppliers whose concessions include more than 600,000 hectares within conservation forest areas across Indonesia.

APP began using our Forest Alert Service in 2016, which utilizes the RADARSAT-2 satellite to “see through” clouds and precipitation, detecting subtle forest disturbances in an area as small as 0.5 hectares. Every 24 days, the system monitors approximately 3.8 million hectares, which also comprise APP's pulpwood suppliers and the Giam Siak Kecil-Bukit Batu Biosphere Reserve.

By monitoring forest cover loss in near real time, APP's pulpwood suppliers can respond rapidly to detected forest changes in production and conservation areas. This has proved particularly helpful in spotting and responding to changes in hard-to-reach forested areas that are normally difficult to monitor from the ground.

Since the monitoring service was introduced, APP reported that losses of natural forest cover have dropped from between 5%-6% to 0.14% in the conservation areas within forest concessions managed by APP’s suppliers.

The possibilities

Earth observation satellites can detect changes in land cover, monitor water quality, identify pollutants, evaluate the health of a coral reef, and help assess how coastal zones are influenced by sea-level rise. With the advent of cloud computing and AI, these satellites are producing petabytes of analyzable data on a daily basis. Whereas previously it took an imagery analyst to meaningfully interpret that information, machines are unlocking insights at unprecedented scales, enabling much more complex research and pinpointing the data and situations that require rapid or more in-depth attention.

Our Earth needs radical solutions today. The strengths of AI — including automating routine tasks, analyzing big data, and highlighting intersections of signal — can become the societal benefits of AI when coupled with satellite imagery and an intention to innovate.

Technological advances, such as machine learning combined with big data, offer astounding possibilities for improving lives. Sharing data and analytical techniques to use data will alter how we identify and respond to natural disasters, epidemics, and the needs of the 1.3 billion people living in extreme poverty.

Realizing these possibilities requires innovative business models that promote cooperation between the public and private sectors. Across the globe, companies, governments, academics, and nongovernmental organizations are exploring how to share data and technological advances — including software and analytic capabilities — and work together to create accurate global maps for humanitarian purposes.

We don’t lack ideas for how satellites and AI could transform development and humanitarian work. These technologies are an opportunity to take a giant leap in our sustainable development progress, but only if we can also establish innovative business models to encourage the private sector and others to focus on putting their data talents to work for the SDGs. Otherwise, the best talent and most resources will continue to focus on AI for autonomous vehicles, financial markets, gaming, and other commercial pursuits. The real payoff, however, is in tackling the challenges presented by the SDGs for the benefit of all humanity and the planet on which we depend.

For more information on how satellite imagery and AI can help you address our world’s biggest global challenges, please click here.