Humanitarian OpenStreetMap Team’s Eastern and Southern Africa (ESA) Hub, USA for UNHCR, and Microsoft’s AI for Good Lab aimed to map Kenya’s Kakuma refugee camp and Kalobeyei settlement, with a vision to create detailed maps that would not only serve humanitarian purposes but also bolster sustainable economic development in the region.Using drones and AI-powered feature recognition, the project produced high-resolution maps vital for humanitarian efforts and sustainable economic development. These detailed maps enable better decision-making, resource allocation, and enhanced living conditions for residents, demonstrating the transformative power of innovative technology in addressing community challenges.

Objective

The project not only helped create map data for immediate humanitarian needs but also provided the basis for long-term planning and development in Kakuma and Kalobeyei, transforming these communities through innovation and collaboration.

The project collected and validated high-resolution aerial imagery of Kakuma Refugee Camp and Kalobeyei Settlement to create accurate, actionable map data. HOT’s work rests on community engagement, and the project engaged and trained local stakeholders, including refugees and internally displaced persons (IDPs), in the data collection process. This ensured inclusivity and empowerment while fostering local ownership of the data.

To enhance capacity building, the project equips UNHCR staff, NGOs, and community members with the necessary skills to effectively use, interpret, and manage map data. This training supports improved decision-making and efficient camp management, strengthening the ability of stakeholders to address challenges within these settlements. Modern technologies, including Artificial Intelligence and Machine Learning (AI/ML), were leveraged to extract, analyze, and update the collected data. These advanced tools provide accurate insights that inform humanitarian efforts, and resource allocation.

Approach

The project aimed to collect accurate, efficient, and reliable geographic data through aerial mapping. A strategic approach was employed to ensure high precision and practicality, focusing on equipment selection, data collection protocols, and analysis methods. Planning was a crucial element, addressing logistical challenges, stakeholder engagement, and obtaining necessary approvals from relevant authorities. Stakeholder engagement was broken down into the phase of- educating stakeholders about drone operations, and ensuring the data aligns with the specific needs and objectives of the project.

The Drone Mapping phase was a pivotal component of the project, harnessing the potential of Unmanned Aerial Vehicles (UAVs), specifically the WingtraOne, to capture high-resolution aerial imagery of Kakuma Refugee Camp and Kalobeyei Settlement, enabling efficient coverage of large areas.

In the remote mapping-labeling phase, high-resolution drone imagery was used to create base maps by digitizing key features like buildings, roads, and waterways from the drone-captured images, forming an accurate and detailed feature dataset.

The field mapping captured detailed Level 3 (L3) data for further analysis in the Kakuma and Kalobeyei areas, laying a strong foundation for mapping and analysis efforts.

The project collaborated with Microsoft’s AI for Social Good Lab led to the development of feature detection algorithms for identifying shelter types, solar panels, and power infrastructure. Algorithms were developed to aid the detection of map features on the drone imagery.

The collected data, analyzed through open mapping techniques, created a detailed representation of the areas, supporting strategic planning and improving living conditions. By aligning data collection with the needs of partners and the community, the project ensured targeted interventions that addressed the specific conditions within the camps and settlements.

Outcome and Impact

This drone mapping project represents a significant advancement in leveraging open map technology for humanitarian support. Through the adoption of innovative drone technology and AI-powered feature detection models, the project has gathered detailed, accurate geographical data crucial for effective planning and response. The integration of open mapping tools, coupled with the application of SPHERE guidelines and spatially derived indicators, has potential to enable a more nuanced understanding of the on-ground realities. This could be instrumental in enhancing the efficiency and efficacy of humanitarian interventions, from resource allocation to infrastructure development.

Lessons Learned

The project provided key insights for future initiatives. First, proper planning for quicker drone approvals and local capacity building is essential. Challenges in processing imagery were significant, as delays occurred due to proprietary data formats and limited agreements with the drone operator. Future agreements should include an explicit requirement to ensure that data is provided in open formats to avoid delays and impact on the geospatial accuracy of the final data. Additionally, the budget for data processing was insufficient, requiring donations of server infrastructure. Future budgets should allocate at least $40/km² to ensure adequate processing power. Coordination between the drone operation and processing teams should be improved, and local communities should be trained to both operate drones and process imagery. This would facilitate faster problem detection and resolution during the project. Looking ahead, the project sets a foundation for the broader adoption of drone mapping and open mapping tools in humanitarian contexts, and highlights the importance of collaborative and adaptive approaches. As we move forward, the lessons learned and the successes achieved will guide our efforts in making a tangible difference in the lives of those in need.

Partners

USA for UNHCR, and Microsoft’s AI for Good Lab