AI-Driven Human-Robot Interaction: Introducing Behaviour Trees into Smart Walkers

Abstract

Smart walkers were developed to assist users with residual locomotion capacities providing physical support, health monitoring, and sensorial, cognitive, and navigational assistance. Multimodal human-robot interaction strategies can enable higher levels of comfort and safety when using the smart walker. Therefore, the functioning of smart walkers must be explainable and transparent, thus reflecting on the acceptance and involvement of its users. Behaviour trees are high-level language adopted in robotics due to its modularity and inherent reactivity, since its architecture facilitates the inclusion of new behaviours seamlessly.This paper proposes the use of artificial intelligence in the core of human-robot interaction strategies with focus on smart walkers to guarantee user’s safety while providing locomotion assistance. The proposed approach was experimentally validated and the results indicate that the implemented behaviour tree correctly predicted unsafe scenarios whereas responding to user’s commands. Such results motivate further studies upon the use of behavior trees for human-robot interaction, a research field still largely unexplored.