Authority-Sharing Control of Assistive Robotic Walkers

Andreetto, Marco (2019) Authority-Sharing Control of Assistive Robotic Walkers. PhD thesis, University of Trento.

[img]PDF (PhD Thesis of Marco Andreetto) - Doctoral Thesis
Restricted to Repository staff only until 01 January 2021.

26Mb
[img]PDF - Disclaimer
Restricted to Repository staff only until 9999.

875Kb

Abstract

A recognized consequence of population aging is a reduced level of mobility, which undermines the life quality of several senior citizens. A promising solution is represented by assisitive robotic walkers, combining the benefits of standard walkers (improved stability and physical support) with sensing and computing ability to guarantee cognitive support. In this context, classical robot control strategies designed for fully autonomous systems (such as fully autonomous vehicles, where the user is excluded from the loop) are clearly not suitable, since the user’s residual abilities must be exploited and practiced. Conversely, to guarantee safety even in the presence of user’s cognitive deficits, the responsibility of controlling the vehicle motion cannot be entirely left to the assisted person. The authority-sharing paradigm, where the control authority, i.e., the capability of controlling the vehicle motion, is shared between the human user and the control system, is a promising solution to this problem. This research develops control strategies for assistive robotic walkers based on authority-sharing: this way, we ensure that the walker provides the user only the help he/she needs for safe navigation. For instance, if the user requires just physical support to reach the restrooms, the robot acts as a standard rollator; however, if the user’s cognitive abilities are limited (e.g., the user does not remember where the restrooms are, or he/she does not recognize obstacles on the path), the robot also drives the user towards the proper corridors, by planning and following a safe path to the restrooms. The authority is allocated on the basis of an error metric, quantifying the distance between the current vehicle heading and the desired movement direction to perform the task. If the user is safely performing the task, he/she is endowed with control authority, so that his/her residual abilities are exploited. Conversely, if the user is not capable of safely solving the task (for instance, he/is going to collide with an obstacle), the robot intervenes by partially or totally taking the control authority to help the user and ensure his/her safety (for instance, avoiding the collision). We provide detailed control design and theoretical and simulative analyses of the proposed strategies. Moreover, extensive experimental validation shows that authority-sharing is a successful approach to guide a senior citizen, providing both comfort and safety. The most promising solutions include the use of haptic systems to suggest the user a proper behavior, and the modification of the perceived physical interaction of the user with the robot to gradually share the control authority using a variable stiffness vehicle handling.

Item Type:Doctoral Thesis (PhD)
Doctoral School:Information and Communication Technology
PhD Cycle:31
Subjects:Area 09 - Ingegneria industriale e dell'informazione
Repository Staff approval on:04 Jun 2019 09:22

Repository Staff Only: item control page