Comprehensive Data Ingestion for Fire Detection

This is the comprehensive intake mechanism for all information vital to PyroSense's platform. Its purpose is to gather real-time data, from multiple origins, ensuring the system has the input needed for accurate analysis and effective decision-making. 

PyroSense integrates an agnostic and highly compatible array of data:

  1. Environmental IoT Sensors are strategically deployed, and continuously collect real-time CO2, temp. and humidity data. They are agnostic in type and protocol, compatible with MQTT, LoRa, Sigfox, and NBIoT, ensuring broad integration. For efficiency, they feature long-lasting batteries (up to 10 years), minimising maintenance.  

  2. Fixed cameras and drones capture high-resolution images and live video. Integrated Vision AI processes this visual data in real-time to detect anomalies like smoke or fire. 

  3. PyroSense gathers data from local weather stations and satellites. Combining granular local data with broad satellite coverage provides a comprehensive understanding of current weather.

  4. GIS provides foundational spatial information, including maps of terrain, vegetation,  infrastructure, etc. 

  5. Firemen Wearables monitor real-time biometrics. AI enhances data for risk pattern recognition, of fatigue or heat stress. Real-time alerts are sent to nearby teams or control centers, enabling proactive intervention.

  • Reliable Sensor Deployment: Sensors should be strategically placed, well-installed, ensuring continuous data collection and security.
  • Data Stream Integration: Integrating data from various sensors, cameras, drones, and meteorological sources is crucial for situational awareness.
  • Data Quality and Calibration: Ensure all data sources are calibrated and high quality to avoid false alarms.  
  • Secure Data Transmission: A strong communication is vital for secure, low-latency data transfer from remote locations.

The diversity and agnosticism of data sources are critical for comprehensive and resilient fire detection. Relying on a single type of sensor or communication protocol creates vulnerabilities. The ability to integrate data from various IoT sensors, visual feeds (cameras, drones), meteorological data, and even human biometrics provides a robust, multi-layered detection system that significantly reduces false positives and increases detection accuracy.

  • The platform must be software and hardware agnostic.
  • Cybersecurity and intercommunication are crucial.

A significant challenge was ensuring seamless interoperability between different sensor types and communication protocols (e.g., MQTT, LoRa, Sigfox, NBIoT) from various manufacturers. As well as, maintaining connectivity in remote, terrains for all sensor types was also an ongoing effort, despite long battery life.

  • Design your system to be compatible with multiple IoT communication protocols from the outset. 
  • Develop algorithms for data validation and fusion to cross-reference information from disparate sources.
  • Consider hybrid communication solutions (e.g., satellite for remote areas)
Sensors and Weather Data
West and South Europe
Panagiotis
Apostolopoulos
Comprehensive Data Ingestion for Fire Detection
Spatial Intelligence for Wildfire Management
Stakeholder Communication & Wildfire Awareness
Core Technologies & Supporting Infrastructure
Protecting Ecosystems Through Fire Prevention Technology
Sensors and Weather Data
West and South Europe
Panagiotis
Apostolopoulos
Comprehensive Data Ingestion for Fire Detection
Spatial Intelligence for Wildfire Management
Stakeholder Communication & Wildfire Awareness
Core Technologies & Supporting Infrastructure
Protecting Ecosystems Through Fire Prevention Technology
Plant Propagation: increased efficiency with improved collecting techniques

Once plants have been collected, they are transferred to our conservation nursery for propagation, or to our seed lab for viability testing and storage. We are seeing increased effectiveness of these methods with freshly collected seeds and cuttings making it quickly to our staff. As many of these individual plants were not previously known, these actions boost the genetic diversity of ex-situ collections, providing a safe place in the face of environmental degradation.

Previously, botanists would need to scale the remote cliff environments where these species occur, making conservation collections difficult and time-consuming to collect and transfer back to nursery staff for propagation. With the Mamba mechanism, collections are quickly collected and transferred to the nursery. 

Fresh cuttings and seeds have a higher success rate in propagation.

 

Drone Collection: Using a drone-based robotic arm to collect inaccessible plants

The Mamba tool allows us to collect plant material via seeds or cuttings from endangered species that we have identified and mapped in the previous building block. This tool has an effective range well over 1000m, making even the most inaccessible areas available for management actions. 

The development of this tool by experienced robotics engineers, expedited the conservation of many species by field staff at the National Tropical Botanical Garden and partners at the Plant Extinction Prevention Program. The Mamba has an interchangeable head system that provides customizable collecting depending on the target species and the type of material necessary for conservation. Many of the components of this mechanism are 3D-printed, which is cost-effective and flexible for speedy development processes. The Mamba is built with readily available drone components which also reduces the cost and building time. The development of this tool was undertaken by P.h.D students, and integrates state of the art hardware and software solutions specifically designed for this application.

When undertaking a project of this type, it is critical to have the proper pairing of experienced field staff with professional robotics engineers, as both parties provide crucial information to guide both development and effective conservation considerations. It is worth noting that the development process was iterative, leaving space for testing and revising the design, and ultimately allowing for deployment of a well-functioning and highly useful tool. 

Drone Survey: location, mapping, and inventory of remote plant populations

Drone tools have been instrumental as a first step in the assessment of cliff floras. Using drones to get unique viewpoints of these environments, we can now map the distribution and abundance of critically endangered endemic cliff species and expedite their conservation. Field surveys have been conducted in Hawaii, the Republic of Palau, and Madeira (Portugal) with extremely positive results.

As drone technology has improved and progressed, this survey methodology has become accessible to a range of conservation practitioners. High-resolution camera sensors allow the identification of a range of plants, from large trees to small herbaceous organisms. Drone pilots can now expect to conduct up to 45 minutes of survey time in a single flight due to increased battery capacity. Usability improvements from software refinements make drones safe and effecient for beginners to use, increasing the uptake of this technology by conservation practitioners.  Most importantly, as drones have become more widely available, the associated costs have been reduced, making them an amazing tool for a range of applications  

Drone are effective tools for the location and inventory of critically endangered species, especially in difficult-to-access environments like cliffs or tree canopies.  Assessment of cliff habitats will be critical to species conservation in these areas, as baseline knowledge of where species occur can guide conservation actions, and help prioritize landscape protection.

Rare cliff endemic and critically endangered Lysimachia iniki being collected by the Mamba sampling arm at its type locality on Kauai
East and South Africa
West and South Europe
Oceania
Ben
Nyberg
Drone Survey: location, mapping, and inventory of remote plant populations
Drone Collection: Using a drone-based robotic arm to collect inaccessible plants
Plant Propagation: increased efficiency with improved collecting techniques
Rare cliff endemic and critically endangered Lysimachia iniki being collected by the Mamba sampling arm at its type locality on Kauai
East and South Africa
West and South Europe
Oceania
Ben
Nyberg
Drone Survey: location, mapping, and inventory of remote plant populations
Drone Collection: Using a drone-based robotic arm to collect inaccessible plants
Plant Propagation: increased efficiency with improved collecting techniques
Education tools

Amphibians are more threatened and are declining more rapidly than either birds or mammals. Amphibian populations are decreasing due to multiple factors, such as climate change, the chytrid fungus, and other anthropogenic factors such as species trafficking. However, the level of threat to amphibians is undoubtedly underestimated because 1294 species (22.5%) are too poorly known to assess, as compared with only 78 birds (0.8%) (Stuart et al., 2004). 

This knowledge deficit underscores the vital importance of educational tools like Ribbit in democratizing scientific research. By lowering barriers to ecological monitoring, apps like Ribbit transform passive observers into active conservation participants. Educational technologies enable citizen scientists to directly contribute to understanding and protecting vulnerable ecosystems, addressing critical research limitations through expanded data collection in under-researched regions.

These innovative platforms increase public awareness about biodiversity challenges while providing accessible pathways for scientific engagement. Unlike bird-focused apps with well-established research infrastructures, anuran conservation has lacked comprehensive citizen science platforms. Ribbit fills this critical gap by empowering individuals to become crucial contributors to amphibian research, turning the tide on data deficiency and supporting global conservation efforts through collaborative, technology-enabled environmental stewardship. It is the first application to include information about over 800 amphibian species, in four languages, including call type, photo, CITES information (whether species are trafficked or used for commercial purposes, addressing GBF targets 5 and 9), IUCN status (whether species are endangered, addressing GBF target 4) and general information on animal behavior and reproduction. 

  • Subject matter expertise: one of our team members (Juliana Gómez Consuegra) worked closely with other experts who were researching the chytrid fungus. 
  • Creating accessible web app: intuitive design of web app allows less-experienced observers to participate and learn.

While the goal is to educate nature enthusiasts, we want to avoid the increase of species trafficking. For this reason, we decided not to allow users to have access to each other's data. That way, an endangered species' location won't be visible to traffickers, on the app. Users only have access to their own data. Once data is shared with GBIF, the data is obscured, so that neither the frog's nor the user's precise location will be disclosed to the general public. This way, we are ensuring that our application is environmentally responsible. 

Mitigate biodiversity loss

Conserving ecosystems is key to curbing climate change, and maintaining ecosystem services (GBF target 11), which are closely linked to over 50% of the world’s GDP. Over 1 million species face the threat of extinction this century; however, selecting which areas to conserve is challenging with the existing data gap, which is biased towards observations in the global north. Increasing the amount of biodiversity data in the Global South is critical in the conservation of endangered species, found at high density in biodiversity hotspots in the Global South. Amphibians are ideal for acoustic identification due to their diverse vocalizations and are crucial ecosystem indicators (Estes-Zumpf et al., 2022), with over 40% of species at risk of extinction (Cañas et al., 2023). Increasing labeled data for the more than 7,000 amphibian species worldwide would enhance conservation efforts and reduce knowledge gaps in vulnerable ecosystems. By using a citizen science platform to aid in the mitigation of biodiversity loss, we help establish local environmental stewardship of these critical habitats (GBF Target 20).

Other citizen apps have shown the potential that citizen science has on mitigating biodiversity loss. eBird, the largest citizen science project related to biodiversity, has 100 million bird observations from users around the world. These observations help to "document the distribution, abundance, habitat use and bird trends through collected species list, within a simple scientific framework." (Sánchez-Clavijo et. al., 2024).  

iNaturalist, another citizen science app, that uses computer vision algorithms for species identification, has also proven successful in mitigating biodiversity loss. To date, the app has over 200,000,000 observations, with 6 million observations per month, globally. On iNaturalist, research-grade observations are shared with GBIF, which in turn uses that knowledge for policy decisions, research, and community building (GBIF, 2023). 

Currently, our app identifies 71 species of frogs and toads, worldwide. Though many of them are identified as least concern (LC) under the IUCN, we do have one IUCN endangered species, the Southern Bell Frog (Ranoidea raniformis). This lack of threatened species included, underscores the need for diverse practitioners to participate in bioacoustic ecological monitoring. Increasing data points on vulnerable species can serve to inform policy decisions using data-driven insights. Local communities and Indigenous Peoples will be a key asset in increasing the number of species included in the app, as their local knowledge allows us to track species in remote regions. 

  • Closing data gaps: get more data from citizen scientists, especially from local communities and Indigenous Peoples.
  • Enabling environmental stewardship: accessibility to a diverse set of users.

We initially set a goal to decrease data gaps in the Global South. However, getting access to enough calls for rare, cryptic, and endangered species in the Global South to train our model proved to be challenging. Therefore, to improve model performance, we turned our attention to as many species as we could tackle, worldwide. Getting users engaged worldwide will lead to more recordings in data-poor regions like the Global South, allowing us to retrain our model in the future with increased data on endangered, rare, and cryptic species. 

This user engagement perfectly aligns with multiple targets, the most evident one being GBF target 20: Strengthen Capacity-Building, Technology Transfer, and Scientific and Technical Cooperation for Biodiversity. But other targets are key in this building block: by increasing the data points, we will be able to identify invasive alien species, addressing GBF Target 6, as well as protecting wild species from illegal trade, by obscuring their location from users. This is aligned with GBF Target 5, which seeks to "Ensure Sustainable, Safe and Legal Harvesting and Trade of Wild Species."