interação multimodal humano-robô-ambiente para a marcha assistida por andador robótico

Abstract

This work presents a strategy of multimodal human-robot-environment interaction for walking assisted by a robotic walker. The strategy aims to meet the needs of users with unbalanced or inconsistent support and interaction forces, in addition to allowing safe navigation and social interaction with other individuals in the environment. Is presented here the Smart Walker Ufes CloudWalker, all component subsystems, in addition to a brief introduction to the mechanical design developed. The Human-Robot-Environment strategy is based on four communication channels: a visual channel, processing the video acquired by a camera, that points towards the user’s face; another visual channel, which processes the video acquired by a camera, positioned on the LRF (Laser Range Finder) sensor, facing the environment ahead, in order to insert context into the scene; an active range channel, based on the data measured by an LRF sensor directed to the user’s lower limbs and another active range channel, based on the data acquired by another LRF sensor, positioned in front of the walker, which scans the environment. An accuracy of 87,5 % in the classification of objects on the scene and the good maneuverability of the SW, using the proposed strategy, maintaining an average speed of 0.25±0.05m/s, even in curves with right angles point to the feasibility of using this interaction strategy in the Ufes CloudWalker