Study Chiang Mai, Thailand, December 23, 2025 – A new evaluation system integrates human-led behavior annotation with canine physiological and movement data to support more objective guide dog assessment.
Guide dog training programs require substantial financial and time investment, and roughly 40% of dogs are released before graduation due to behavioral, medical, or suitability concerns. Because resources are limited and demand for guide dogs continues to rise, schools must identify the most promising candidates early in the training pipeline. Historically, expert staff rely on subjective evaluations that depend heavily on human experience, intuition, and long-term familiarity with canine behavior.
This case study presents a system designed to support a more objective and data-driven approach. Developed in collaboration with Guiding Eyes for the Blind, the system enables a human operator to annotate canine behavior in real time while simultaneously capturing physiological and accelerometer data from the dog. The integration of these modalities allows evaluators to connect observable behaviors with underlying physiological states and movement patterns, potentially revealing indicators of guide dog success that humans might overlook.
Importantly, the system was designed from the ground up to meet the needs of both humans and dogs. From the canine perspective, comfort, non-invasiveness, and freedom of movement were treated as essential design criteria. For human evaluators, the interface had to be intuitive, efficient, and usable in dynamic training environments. This dual-user design philosophy ensures the resulting system is practical, humane, and aligned with real-world training workflows.
As a first step toward more standardized and quantitative guide dog evaluation, this system offers meaningful potential. By reducing subjectivity and augmenting expert judgment with continuous sensor-based insight, guide dog schools may ultimately improve placement rates, reduce training costs, and identify behavioral or stress-related concerns earlier in development.
Source: Mealin, S. P., Foster, M., Roberts, D., et al. (2017). Creating an Evaluation System for Future Guide Dogs: A Case Study of Designing for Both Human and Canine Needs. International Conference on Animal-Computer Interaction. No DOI was provided in the supplied text.
