Drones, AI and the future of wildlife conservation

Episode Summary

Executive Summary: This episode examines how drones, computer vision, and AI are transforming wildlife conservation, especially through EU-funded Wild Drone and Wildbotics. Interviews with ecologist Blair Costello, PhD researcher Camille Rondeau-Saint-Jean, and Ol Pejeta’s William Jiroge show technology improving monitoring, safety, and data quality while still relying on rangers, local knowledge, and ethical oversight.

Main Topics: Drones as a new tool for wildlife ecology (Priority: 5/5): Blair Costello explains moving from manual observation to aerial drone imagery to study collective animal behavior in the field, especially for zebras, antelope, and buffalo. Balancing data collection with animal disturbance (Priority: 5/5): The conversation emphasizes that drones can disturb wildlife differently by species, altitude, habitat, and group composition, so pilots must test methods carefully and minimize noise. Computer vision and machine learning for conservation data (Priority: 5/5): The episode highlights automated analysis for camera traps, individual identification, behavioral patterns, and bioacoustics, with the caveat that model performance depends on training data quality. Photogrammetry for measuring marine and terrestrial animals (Priority: 4/5): Camille describes using off-the-shelf drones and photogrammetry to measure harbor porpoises and rhinos, extracting body size and condition from high-resolution aerial images. Conservation operations at Ol Pejeta Conservancy (Priority: 4/5): William Jiroge outlines how drones could assist rhino monitoring, infrastructure checks, and security, while fitting into a larger conservation-tech lab and stakeholder framework. Human expertise, rangers, and collaboration remain central (Priority: 5/5): All speakers stress that technology complements rather than replaces field staff; rangers, pilots, ecologists, and computer scientists must work together for effective conservation. Open science, training, and international exchange (Priority: 4/5): Wild Drone’s EU-supported structure encourages open-source datasets, regular in-person meetings, hackathons in Kenya, and cross-border PhD collaboration.

Key Arguments: Manual field observation remains foundational, but it is too limited for continuous group-level behavioral monitoring; drones provide the aerial perspective needed for new ecological questions. Drone use in conservation must be species-specific and ethically tested because disturbance responses vary widely; buffalo, zebras, and impala reacted very differently. Automation reduces human error in repetitive tasks such as multi-drone handoffs, image annotation, and filtering camera-trap blanks, freeing researchers to focus on interpretation. AI is most useful when trained on large, high-quality ecological datasets; model accuracy falls when data are sparse or unrepresentative, so training and checking remain essential. Photogrammetry can turn drone imagery into precise morphometric data, enabling measurements of body condition, pregnancy, calf age, and health from above. At Ol Pejeta, drones could improve rhino census accuracy, reduce risk to rangers, and support infrastructure and security monitoring, but flights must be regulated and coordinated. Conservation technology should augment, not replace, human roles; rangers and local knowledge are indispensable to fieldwork and decision-making. Open-source datasets and collaboration between ecologists and computer scientists are critical to building tools that actually work in conservation settings.

Data Points: Wild Drone PhD researchers: 12 - International researchers in the Wild Drone network EU-funded doctoral training networks: 2 - Wild Drone and Wildbotics Project timing: Wild Drone is coming to an end - The network is concluding as Wildbotics expands the approach Species studied in Blair Costello’s drone work: 4 - African buffalo, two zebra species, and impala Drone relay altitude difference: 10 meters - In Blair’s early work, the second drone was sent up 10 meters above the first during handoff Drone teams at Ol Pejeta: 5 drone teams - Mentioned in a cited CIO Africa article about the program Drone missions at Ol Pejeta: 260 missions - Reported in the same article as part of the Wild Drone output Drone flights at Ol Pejeta: 60 drone flights - Reported in the same article as part of the Wild Drone output Rhinos at Ol Pejeta: about 250 - William describes the conservancy’s closely monitored rhino population Black rhinos at Ol Pejeta: 1,8089 (spoken as 18089) - William states the conservancy is the largest sanctuary of black rhino in Eastern/Central Africa Northern white rhinos left in the world: 2 - Najin and Fatu are the last known northern white rhinos Rhoinos not poached at Ol Pejeta: 8 years - William says there has been no rhino poaching event there for eight years Community supported by the conservancy: more than 50,000 people - William links conservation to local benefits and governance Operations funded by tourism: 70% - William says conservancy operations are largely funded by visitors Rhino monitoring frequency target: at least every 4 days - Rangers aim to account for every rhino on this schedule Harbor porpoise length: 1.35 to 1.8 meters - Camille describes the approximate adult size of the species she studies Porpoise field record: 9-10 years - A distinctive female porpoise returned for multiple summers and had been observed over nearly a decade Student countries in Wild Drone: 12 different countries - The network brings together researchers from Europe and beyond Musicology and ecology transition: Returned to high school to redo science classes - Camille explains her path from music studies to ecology and bioacoustics

Pivotal Quotes: "I see basically these tools as allowing us to do more in terms of collecting more data at higher resolution that lets us, you know, then bring our human insight into interpreting that data." — Blair Costello: Why drones and AI complement rather than replace scientists "It's not just a research program... the most important bit is linking academic to industry, in this case conservation, bringing academia to tackle challenges that are faced directly in conservation." — William Jiroge: Why Wild Drone matters at Ol Pejeta "I feel like right now I've done it all by hand, but I have a fantastic data set for training." — Camille Rondeau-Saint-Jean: Current limits and future promise of automating drone-based measurements

Implications: The episode suggests conservation is shifting toward data-rich, cross-disciplinary, and partly automated workflows. Future gains will depend on better datasets, safer drone design, and strong collaboration with rangers and local institutions.