Wearable interfaces for symbolic communication by working dogs

Abstract

The objective of this dissertation is to develop wearable systems that allow working dogs to communicate accurately with humans. For example, a guide dog could generate alerts telling the human to either 'wait' or 'go around’ a given obstacle, which currently they cannot do. Working dogs in search and rescue could communicate critical information beyond line of sight and hearing. Finally, during a seizure, medical alert dogs could request help from nearby humans. This work describes two types of wearable systems capable of addressing the challenges in these scenarios. The first approach relies on wearable interfaces based on biting, tugging, and touching that working dogs can use to generate these alerts. The second approach relies on identifying and detecting gesture movements sensed from a collar-worn device that dogs can learn to perform to generate alerts.