Data Driven Enrichment for Georgia Aquarium’s Sea Otters
Enrichment is a crucial component of animal care in zoos and aquariums, designed to stimulate natural behaviors and improve welfare. Many facilities rely on off-the-shelf dog toys to provide enrichment as they are affordable, durable, and readily available. However, these generic toys are not tailored to the specific needs of different species, limiting their effectiveness. More specialized tools, designed with species-specific behaviors and ecological roles in mind, have the potential to offer greater cognitive engagement and behavioral benefits.
Charles Ramey, a Georgia Tech PhD student in the School of Interactive Computing, studies computer science and intelligent systems at TSRB. His research focuses on animal-computer interaction, leveraging experience in machine learning, rapid prototyping, low-power embedded systems, and robotics. In collaboration with Georgia Aquarium’s trainers, staff, and researchers, Ramey develops instrumented enrichment objects—toys embedded with sensors—to better understand how animals interact with enrichment over time. Rather than focusing solely on designing species-specific devices, the project emphasizes using these technologies to quantify engagement, determine when animals may benefit from novel enrichment, and detect behavioral changes that could indicate shifts in health or well-being.
His recent pilot study, Shelling Out the Fun: Quantifying Otter Interactions with Instrumented Enrichment Objects, published in December 2024, utilizes sensors embedded in enrichment objects to characterize how sea otters interact with food-dispensing toys and determine whether researchers can successfully collect quantified interaction data using enrichment devices. The study involved the five sea otters (Enhydra lutris) at the Georgia Aquarium, testing two instrumented enrichment objects: a KONG Genius Leo dog toy, which the sea otters already use as an enrichment device, and a novel, scallop-shaped device. The motion sensors within these devices measured otter engagement with the toys based on the time and effort required for each otter to extract the food rewards within the toys. Results showed that, on average, the otters took longer to get less food from the novel object, which may be due to a learning curve. During the study, data were collected from both embedded sensors and video cameras to track interaction times. The researchers weighed the food in each toy before and after each otter interaction session. Anecdotally, the team noted differences in how the otters interact with the toys through the video footage.
“One otter, Bixby, did not interact with the scallop at all,” says Ramey. “Each interaction session was conducted in an isolated pool to prevent any of the otters from learning from the others. They all did variations of shaking and banging each object against the walls and floor. Also, each otter was only given each object once during the pilot study, so we don’t have any observations over time yet,” he concludes.
The pilot study’s approach to collecting quantified data proved feasible, demonstrating that innovative enrichment tools can capture detailed behavioral data in real-world settings. While the primary focus was on validating the data collection method, the study also observed a coincidental correlation between the novelty of the device and the time it took otters to extract food, suggesting potential avenues for future exploration.
Ramey’s work aligns with animal-computer interaction studies that use wearable electronics to assess health and behavior. His study emphasizes the use of durable, safe, and cost-effective species-specific enrichment tools. The research also highlights the limitations of repurposed dog toys and advocates for the development of purpose-built alternatives using sustainable methods and materials.
This research sets a precedent for developing data-driven enrichment strategies across zoos and aquariums. The Association of Zoos and Aquariums (AZA) establishes enrichment standards, and Ramey’s findings may hopefully contribute to enhancing these guidelines. By quantifying the effectiveness of enrichment, researchers can potentially improve the welfare and cognitive engagement of captive animals. The study also opens possibilities for automated tracking and adaptive enrichment systems, refining how facilities monitor and optimize animal well-being.
Future studies will build on this pilot, refining device design and tracking behavioral trends over extended periods. This approach provides scalable solutions for animal enrichment, thereby enhancing the quality of life in managed care environments.