Restoration is not just about planting—it’s about ensuring long-term impact. Our AI-powered MRV system provides real-time tracking of restoration progress and environmental health. It also addresses critical issues like illegal fishing, poaching, and deforestation, empowering communities to protect their restored ecosystems. This system integrates satellite data, drone imagery, and AI analytics to provide actionable insights, which can be adapted to other restoration or conservation efforts. It also supports transparency and accountability, ensuring stakeholders can measure progress and outcomes effectively.

Our licensing and training platform equips communities to independently build, operate, and maintain drones. This approach is hands-on and collaborative, fostering local expertise and enabling communities to adapt the technology to their unique needs. The platform goes beyond technical skills, creating a foundation for communities to innovate and modify drones for additional applications such as surveillance, mapping, and precision agriculture. Importantly, the platform fosters a feedback loop where communities share their innovations, enriching the broader global network of users.

Our modular drones are designed for accessibility, adaptability, and sustainability. Initially crafted using wooden components with fewer than six screws and zip ties, they are simple to assemble, repair, and replicate using local materials, empowering communities to lead restoration projects independently.

As we’ve advanced, we’ve integrated hydrogen fuel cells and hybrid-electric propulsion systems, enhancing flight endurance, energy efficiency, and environmental sustainability. These innovations enable drones to cover larger areas and operate in remote environments while reducing their carbon footprint.

The modular design ensures flexibility for continuous adaptation, allowing communities to upgrade drones with tools like cameras or sensors for monitoring. This approach combines simplicity and cutting-edge innovation, bridging grassroots empowerment with scalable, impactful environmental restoration.

Waterbird monitoring is the foundation of protection and management strategies for almost types of wetland ecosystems. With the continuous improvement of wetland conservation infrastructure in China, including remote devices for collecting large amounts of acoustic and visual data of wildlife, the demand for data filtering and analysis technology is increasing. Deep learning based object detection has become a fundamental solution for big data analysis and has been tested in multiple application areas. However, these deep learning techniques have not yet been tested for detecting small waterbirds in real-time monitoring videos. We propose an improved detection method that adds additional prediction heads, SimAM attention modules, and continuous frames to YOLOV7, called YOLOv7 Waterbirds, for real-time video surveillance devices to identify attention areas and perform waterbird monitoring tasks (identification, counting, and density estimation). Based on the waterbird dataset, the average accuracy (mAP) value of YOLOv7 waterbird is 67.3%, which is about 5% higher than the benchmark model. In addition, the recall rate of the improved method is 87.9% (accuracy=85%), and the recall rate for small waterbirds (defined as pixels less than 40x40) is 79.1%, indicating that its performance in small object detection is superior to the original method and many other popular deep learning algorithms. This algorithm can be used by protected area management departments or other organizations to use existing surveillance cameras for higher precision monitoring of aquatic plants, which to some extent contributes to wildlife conservation.

Through a series of scientific measures such as micro terrain modification, water level regulation, and wetland ecological restoration, we aim to create a habitat environment that can meet the needs of various migratory birds. After the completion of the restoration project, in daily management, the high tide roost site needs to maintain a certain proportion of bare flats, shallow water areas, deep water areas, and controlled low vegetation areas. By manually controlling the water level to ensure the relative stability of different water level areas, controlling the height of weeds to maintain the bare flats area, in order to provide the habitats of different migratory waterbirds such as shorebirds, herons, gulls, ducks, etc. Using unmanned devices such as video surveillance to assist in monitoring waterbird communities, in order to evaluate the patterns of habitat use by these migratory birds and subsequently assess habitat quality.

Under the application of the trap for intermittent harvest, the best results were achieved with the following combination of variables: maize bran (supplementary feed) x maize bran (trap bait) x O. Shiranus (species) x 2 fish/m2 (stocking density).

The total yields under this combination were 25 percent higher than in the control group with single batch harvest. A higher stocking density (3 fish/ m2) led to a slightly higher total harvest in the control group, but to a lower net profit. The use of pellets reinforced both effects and was the least economical.

Results from the on-farm trials (see Figure 1) have demonstrated the functionality and the excellent catch effect of the traps. Over the three-month on-farm trial period, the trap was used 2 to 3 times a week and a total of 27 times. On average, around 120 small fish – an equivalent of 820 grams – were caught each intermittent harvest. With the use of the trap, all households reported that they now eat fish twice a week. Before that, fish consumption was between one and four times a month.

The benefits:

• Reducing the competition for oxygen and food among the fish in the pond and thus measurable increase in yield.
• Improved household consumption of small, nutritious fish and better cash flow.

Success factors:

• Traps are easy and inexpensive to build (USD 3).
• Traps are easy to use, also for women.
• Directly tangible added value thanks to easy and regular access to fish.

Examples from the field

Overall, the user experience of households engaged in the on-farm trials was very positive:

“As a family we are now able to eat fish twice and sometimes even three times a week as compared to the previous months without the technology when we ate fish only once per month.” (Doud Milambe)

“Catching fish is so simple using the fish trap and even women and children can use it.” (Jacqueline Jarasi)

“It is fast and effective compared with the hook and line method which I used to catch fish for home consumption that could take three to four hours but to catch only three fish and thus not enough for my household size.” (Hassan Jarasi)

• Involvement of the local community
• Responding to community needs 

On-station trials

In a series of experiments conducted at the National Aquaculture Center in Domasi, the project team tested the trap for intermittent harvest with different baits in ponds (200 m2) stocked with different species (Coptodon Rendalli vs. Oreochromis Shiranus) at different densities (1 vs. 2 vs. 3 fish per sqm.). In addition, further tests were carried out to determine the time and intervals it takes to catch a certain amount of fish. As a control and for comparison, additional ponds were stocked with O. Shiranus and C. Rendalli fed with maize bran or pellets for single batch harvest to represent customary forms of rural aquaculture in Malawi.

On-farm trials

At the time when the trap was technically functional, households that wanted to test the trap under every day, real-life conditions were identified. Over three months, six households tested the trap and documented the catch.

The trap is made from wire mesh and shaped like a cylinder. Two additional wire mesh pieces shaped like a cone are attached at both ends. The diameter of the narrower end is kept smaller to allow only small fish to enter the trap. To lure them in, bait is placed inside. A piece of a net holds the bait. A string is fixed to the trap so that users can easily sink and retrieve the trap.

In the context of fisheries and aquaculture, the fish trap represents an evolution of existing harvesting methods. Unlike active fishing gear, such as seines, the fish traps require less labor and energy, which makes them very efficient in terms of catch effort. In addition, the fish traps do not physically harm the caught fish, so the fish can be taken out of the trap alive and in good health. Early experiments on partial harvests in aquaculture in Malawi date back to the 1990s, when different tools for intermittent harvest were tested. However, due to the inefficiency and labor-intensity of the methods, there has been no broad application or further developments.

Based on this knowledge, further literature research, and expert discussions, the idea was born to build and test a size-selective fish trap to regularly harvest the juveniles of the initial fish stock. This innovation is thought to control the stocking density, to optimize the use of supplementary feeds, and to not exceed the carrying capacity of the pond. Ideally, a successful application of the fish trap would result in households increasing their overall aquaculture productivity, whilst harvesting small quantities of small fish much more regularly than has been customary in aquaculture to date. The intermittently harvested fish can be consumed within the household or used to generate small amounts of regular income. Meanwhile, the initial fish stock (parent fish) will be grown to a larger size for the final harvest.