Glasgow Coma Score Estimation Module for Deployment on Mobile Robots in Rescue Missions

This work presents a module to estimate a score according to the Glasgow Coma Scale (GCS) for deployment on mobile robots in order to determine the level of consciousness of victims in mass accidents to support rescue personnel. To determine a score according to the Glasgow Coma Scale multiple different methods are required. Pose estimation in order to find and determine a motor score, facial landmark detection to track eyelid movements in order to determine an eye score and natural language interaction to evaluate the verbal score. Our module is able to determine an eye score of 3–4, verbal score of 4–5 and motor score of 6 due to excluded pain stimuli. This work represents a foundation for an automated estimation of a score according to the GCS which will comprise enhancing pose estimation methods with more diverse training, identifying optimal facial landmark detection methods to find a balance between accuracy and performance. Further, refinement of natural language interaction is aimed for.