Solution for: Changing Unsustainable Trade

An Overview Of Our Solution

Conservation X Labs’ developed Sentinel, a smart camera used to better understand and protect wildlife from poaching threats and illicit trade as well as mitigate human-wildlife interactions. Sentinel addresses the greatest threats on the frontlines of the biodiversity crisis, by providing advanced tools to process information and deliver valuable and actionable insights to rangers to be better equipped in protecting critical wildlife and habitats. Sentinel currently operates on the Osa Peninsula, in Costa Rica, identifying jaguars, tapirs, white-lipped peccaries, and other species of interest to provide rangers with accurate population densities and migration patterns. Sentinel also identifies suspicious human and vehicle operations for rangers to use as a first line of defense against poaching and wildlife trafficking of the biodiversity in these ecosystems.

Who is this solution impacting?

Ecosystems Impacted

Forests

Community Type

Rural

First name

Alex

Last name

Dehgan

Organization type

Nirlaba

Email Address

taylor@conservationxlabs.org

Context Analysis

Today’s trusted and most widely used tools in conservation technology have enabled the capture of wildlife data that would be impossible for humans to collect alone and also fosters a better understanding of the behavior and threats to endangered species or species of interest globally. Many of the most significant underlying drivers of extinction are caused by time-sensitive scenarios such as poaching, invasive species, and zoonotic diseases. It is common for teams to devote significant portions of their resources to manually retrieve data from trail cameras and individually analyze thousands of images. As a result, acting on these insights or using them for decision-making is no longer timely or feasible. There is currently no solution that provides near real-time information from the wild to inform conservation activities. These challenges are particularly difficult for Osa Conservation, CXL’s on-the-ground partner in Costa Rica, in their efforts to monitor and protect endangered wildlife in the Osa Peninsula, a region that is home to over 2.5% of the world’s biodiversity. With over 700 square miles of rich and diverse ecosystems, the Osa Peninsula and the species that thrive there are threatened by numerous activities, not limited to habitat encroachment, poaching, and illicit wildlife trade. Endangered species, in particular jaguars, are most at risk due to illegal wildlife trade in this region, and the loss of species may have dire consequences on the region’s tourism, economy, and culture.

Describe the technical solution you wanted the target audience to adopt

Existing monitoring devices like trail cameras and acoustic sensors function as simple recorders, amassing indiscriminate data that can only be analyzed and interpreted retrospectively. In response to this, CXL created Sentinel, to connect to standard wildlife monitoring tools and immediately process data via local AI models. By only transmitting useful information out of the large amounts of filtered data, the Sentinel enables real-time awareness whenever something critical is detected – like the presence of a poacher – so users can take immediate action. Designed to be affordable and accessible, the Sentinel allows non-technical users such as rangers, conservationists, scientists, farm holders, and others to leverage AI technologies with a powerful platform that does not require coding knowledge or experience. Sentinel makes real-time environmental monitoring possible by placing the power of AI in the hands of those working to protect threatened wildlife and critical ecosystems.

Describe your behavioral intervention.

Prior to partnering with Conservation X Labs, Osa Conservation spent a significant amount of staff time and resources manually checking trail cameras, swapping batteries and SD cards, and clicking through collected photos and videos to identify valuable information. By the time valuable information is identified in this arduous process, the subject has moved locations, the critical event has passed and action is often too late. Osa Conservation is the first organization in Latin America to use Sentinel devices across the rainforest to provide live continual monitoring of the surrounding habitats. Given Sentinel’s ability to provide self-health checks on battery life, memory capacity, and movement if they are stolen, Osa Conservation has helped prove immense value in reducing the amount of time spent retrieving actionable information and is helping to encourage the use of Sentinel with other potential users struggling with the same challenges. Sentinel deployments in the Osa Peninsula jungle enabled their team to spend fewer resources and fuel on travel to trail camera sites and reallocate resources to further the positive impact of their work. With the goal of 100 cameras constantly monitoring 4,500 hectares of rainforest, the anticipated reaction time from identifying threats and events in the wild to action by 1440x.

Behavioral Levers Utilized

Emotional Appeals

Information

Material Incentives

As needed, please explain how you utilized the lever(s) in more detail.

The primary goal of Sentinel is to provide valuable information to users in need of quick data to make actionable decisions and free up limited resources. The ability to reallocate resources and make real-time, sound, data-driven decisions about various conservation issues – the presence of an endangered species, poacher, or illicit trade activity – is valuable for impactful conservation action. Sentinel is a tool that helps identify some of the most endangered species in the region as well as reinforces national rules prohibiting wildlife trafficking, and protects charismatic animals which appeal to the emotions of locals and visitors. Sentinel informs patrols and management decisions and also discourages potential traffickers and criminals from harming wildlife with the added threat of getting caught. The Sentinel is a product that uses a combination of these levers to alter the behavior of how people treat and protect the biodiversity of the Osa Peninsula.

Describe your implementation

The Sentinel is a system comprised of four distinct pieces:

A low-powered universal AI adapter that makes current environmental monitoring tools such as trail cameras capable of running local AI models. Notifications about events in the wild are sent to users in real-time as they are detected and processed;
A user dashboard to manage fleets of Sentinels remotely, deploy new models & updates over the air, and view incoming insights;
An online AI Model Marketplace (akin to an app store) to browse, download, and install models onto the Sentinel that fit the users’ needs; and,
An online pipeline within the Model Marketplace where users input the questions they hope to answer. Our backend then automatically creates a custom AI model for the user to deploy to their Sentinels, without the users touching a single line of code.

By building a full end-to-end solution for organizations to better protect the wildlife in their region, the Sentinel becomes a sustainable tool used daily as a part of their workflow. Sentinel smart cameras are being deployed across the rainforest of the Osa Peninsula, initially to identify jaguars, tapirs, white-lipped peccaries, and suspicious human/vehicle behavior. This allows Osa Conservation and local rangers to have quick access to where these animals are located, and respond quickly to imminent threats of poaching. CXL continues to work closely with the Osa Conservation team to ensure the Sentinel cameras function well in the rainforest ecosystem and provide value and information to their teams.

Throughout deployment, the team has faced a number of challenges, such as switching from cellular network connection to satellite after field testing communication protocols. The team has also improved the battery life of Sentinels as a result of our field testing with the Osa Conservation team. We measure success by the utility of the Sentinel, as we aim to monitor 4,500 hectares of rainforest for real-time wildlife protection.

Describe the leadership for your solution. Who is leading the implementation?

Conservation X Labs is led by co-founders Dr. Alex Dehgan and Dr. Paul Bunje. As the Former Chief Scientist at USAID, Dehgan founded their Office of Science and Technology, and previously established the first National Park in Afghanistan. Bunje has over a decade of executive experience, raising over $80M for Global Scientific Challenges and Prizes during his time at XPRIZE.

Sentinel development and implementation is led by Sam Kelly, with extensive experience in field-ready electronics. He previously co-founded FaunaLabs with Duke University’s Nowacek Lab. Dr. Dante Wasmuht leads the Sentinel’s AI research and model development, with a background as a Wildlife Behavior Neuroscientist from Oxford University. Henrik Cox leads the product management, development, and marketing of the Sentinel. Dr. Chad Gallinat leads the operations for the Sentinel team, and has previously led research on field-deployable devices & managed multi-million dollar international collaborations w

Share some of the key partners or stakeholders engaged in your solution development and implementation.

CXL has two types of partnerships imperative for success: 1) technical partners and 2) deployment partners.

Technical partners:

University of Texas, Austin and the Texas Advanced Computing Center provide compute time for model development and data processing capabilities as part of an effort to use excess compute time for conservation; and,
Edge Impulse transfer heavy cloud-based machine vision models to lighter edge-ready models and agreed to give their design for an all-in-one camera version of the Sentinel device in exchange for helping to commercialize the camera.

Deployment partners:

Island Conservation to identify invasive rodents on the US West Coast and Hawai’i;
Earthranger and SMART to integrate data insight delivery and expand the scalability of such deployments;
Sabine Plattner African Charities (SPAC) to characterize primate behavior and re-identification of gorillas in the Democratic Republic of the Congo; and,
Osa Conservation to detect poaching of endangered species, most notably jaguars, in Costa Rica.

Conservation X Labs has been learning from working with early partners and customers to understand how they would most benefit from working with Sentinel and adjust for more people to use in the future. Insights from Sentinel devices are sent to the online user dashboard, which can be viewed, corrected, and shared with other users easily, or downloaded to share externally.

The Model Marketplace is a community-first platform designed to work with the Sentinel and is built to serve users by making AI technology accessible for on-the-ground conservation programs. The platform allows researchers and developers to host models that may be beneficial to the wider conservationist community to download and implement in their own projects. The Model Marketplace will be open source to allow the community to learn from each other, have access to advanced technologies, and contribute to it over time.

Who adopted the desired behavior(s) and to what degree? Include an explanation of how you measured a change in behavior.

Sentinel significantly impacts the speed and scale of data processing to enable rapid response and relevant action relating to 1) automatically identifying endangered jaguars in Costa Rica and preventing poaching incidents in real-time, 2) detecting and supporting the capture of invasive rodents in California and Hawaii, and 3) understanding and addressing feline leukomyelopathy (FLM), a neurological disease present in Florida panthers. In the case of rodent detection, Conservation X Labs’ partner Island Conservation used models to reduce data processing speeds by over 90% and reduce data processing costs by 4x, freeing up resources. The Florida Wildlife Commission used Sentinel models to identify FLM in Florida panthers with 90% accuracy to better understand and respond to the spread of this neurological disease. The impact of the reallocation of resources has yet to be identified but CXL anticipates the result to be significant.

How did you impact the environment (biodiversity conservation, ecosystems, etc.)? Please be specific and include measurement methodology where relevant.

CXL’s application of AI is critical to reduce the amount of data transmitted from remote locations and efficient at filtering large image datasets while highlighting events that are of interest to a particular question. Most devices rely on low-power satellite systems or LoRa/SigFox protocols that are only capable of sending short text messages so it is necessary to process the data on the device, effectively compressing and prioritizing the data for the end users need most. Automatically processing the data with AI can cut time by over 90% and reduce cost of data processing by 3-5x. Depending on the available training data, the algorithms vary between 80-95% accuracy which matches human performance on some datasets. This provides conservation organizations with instant threat alerts. By monitoring the rainforest with Sentinels, we are able to provide the fastest and most accurate information to use for the protection of remaining jaguars, and other threatened species in the area.

How has your solution impacted equity challenges (including race, ethnicity, social class/income, indigenous communities, or others)?

Conservation X Labs prioritizes working with local communities, organizations, and people who best understand the needs of the region in deployments and use of the Sentinel and will continue to do so. Due to the area of impact, this includes underrepresented populations such as indigenous communities and local people. For each deployment or program, the Sentinel team secures powerful local partnerships that give a unique and relevant perspective and work with experts in the field. In working with Osa Conservation, a local field-based conservation organization, the greatest impact is when it is in the hands of users who best understand the problem. In rural and remote locations, that is often the most underrepresented populations around the world.

What were some sustainable development co-benefits?

While specific impacts vary, the basic metrics of success are: 1) SDG 15.7 “Eliminate poaching and trafficking of protected species” Sentinel enables direct interventions outside of cellular networks to protect species and help ecologists better understand wildlife behavior, and 2) SDG 15.9 “Integrate ecosystem and biodiversity in governmental planning” Sentinel provides a continuous monitoring network enabling dynamic decisions to inform effective management in the most biodiverse places.

Sustainability: Describe the economic sustainability of your solution.

The development of the Sentinel as a product has relied substantially on grant funding to date. We have also received funding through competitions, awards, philanthropy, and contracts. We are beginning to manufacture Sentinels in small batch processes and are securing additional customers to grow our market-based revenue. However, we are still dependent on additional funding through grants until we can scale to a point where we can be self-sufficient and economically sustainable.

Return on investment: How much did it cost to implement these activities? How do your results above compare to this investment?

The cost of development and deployment for our work with Osa Conservation has exceeded USD$55,000 after an initial fund of USD$50,000 from Osa Conservation’s associated partners. As one of our first research deployments of the Sentinel, we anticipated our costs would exceed the funding provided for the project. However, since our deployment with Osa Conservation in May 2022, and projected second deployment in March 2023, the Sentinel devices have become far more economical in cost per unit for manufacturing. Future deployments and projects using the Sentinel will yield far greater returns to investment in comparison to our initial fieldwork with Osa Conservation in Costa Rica.

How could we successfully replicate this solution in Latin America?

This solution is easily replicable across Latin America. Sentinel smart cameras can be purchased and deployed in the wild, even in locations with no cell service, given our ability to transmit data over satellite affordably. Custom AI models can be bought for training and deployment for specific species of interest across Latin America as well.

The funding required depends on the size of project deployments. Our Sentinels retail at USD$350 for cellular units and USD$400 for satellite versions, with a monthly subscription per device of USD$20 for cellular and USD$35 for satellite units. A custom model retails for USD$5,000 apiece. For reference, we are preparing for field deployment in Colombia in the coming months for 40 Sentinels and 2 AI models, with funding of USD$42,000.

Wilayah Peru Berlaku

Loreto

Madre de Dios

Jelaskan bagaimana solusi teknis menangani perdagangan satwa liar di Amazon Peru.

CXL developed the Sentinel, an AI toolkit for the frontlines of the biodiversity crisis. Sentinel has two primary components: 1) a hardware device that replaces or connects to standard wildlife monitoring tools and immediately processes data via local-AI models, and 2) a machine learning platform used to leverage existing AI models and train custom models without understanding code. The device filters through large amounts of data to transmit only useful information, enabling real-time identification so that users can act immediately (e.g. real-time alert of poaching or species). Designed to be affordable and accessible, the Sentinel online dashboard provides users with clear data visualization of their received insights, and gives rangers in any location instant information on the whereabouts of endangered wildlife at risk of trafficking, and suspicious signs of poaching activity to respond to.

Jelaskan keberlanjutan dan skalabilitas solusi teknis Anda

CXL plans to scale the product directly. The proposed work will help lower the cost of Sentinels to reach a wider audience. More devices means more eyes in the field sending insights to users about time-sensitive scenarios and protecting more species. CXL also intends to integrate Sentinel with other major platforms and technologies to offer a better product to users. For example:

CXL integrates with EarthRanger and SMART platforms; which are used as centralization software to monitor techno

Jika ada, jelaskan tingkat kemajuan solusi

The solution is already being used in Global South countries and is ready for additional deployments with a business model developed. The team continuously performs various deployments and field tests to ensure the robustness of this product. Sentinel has had a significant impact on the speed and scale of data processing to enable subsequent action relating to 1) automatically identifying endangered jaguars in Costa Rica and preventing poaching incidents in real-time, 2) detecting and supporting the capture of invasive rodents in California and Hawaii, and 3) understanding and addressing feline leukomyelopathy (FLM), a neurological disease present in Florida panthers. In the case of rodent detection, CXL’s partner Island Conservation used models to reduce data processing speeds (in terms of human hours) by over 90% and reduce data processing costs by 4x. The Florida Wildlife Commission used Sentinel models to identify FLM in Florida panthers with 90% accuracy. The impact of the reallocation of resources has yet to be measured or identified but CXL anticipates the result to be significant.

Apa sumber pendanaan tambahan untuk solusi?

The development of Sentinel as a product has relied substantially on grant funding to date. We have also received funding through competitions, awards, philanthropy, and contracts. We are beginning to manufacture Sentinels in small batch processes and are securing additional customers to grow our market-based revenue. However, we are still dependent on additional funding through grants until we can scale to a point where we can be self-sufficient and economically sustainable.

CXL plans to scale the product directly. The proposed work will help lower the cost of Sentinels to reach a wider audience. More devices mean more eyes in the field sending insights to users about time-sensitive scenarios and protecting more species.

Biaya Solusi Inovatif

A Sentinel deployment includes hardware, calibration, and other software set-up and training and can be scaled up or down depending on customers specific needs. An approximate budget follows, but would need to be customized based on specific requirements

Budget:

Inventory & Supplies

$35,000

Travel & expenses

$10,000

Misc. & Overhead

$5,000

Total Approximate Budget

$50,000

Replication and Scale

Conservation X Labs

Video file