Development and Validation of a Robust Speech Interface for Improved Human-Robot Interaction

Abstract

Robotics technology has made progress on a number of important issues in the last decade. However many challenges remain when it comes to the development of systems for human-robot interaction. This paper presents a case study featuring a robust dialogue interface for humanrobot communication onboard an intelligent wheelchair. Underlying this interface is a sophisticated software architecture which allows the chair to perform real-time, robust tracking of the dialogue state, as well as select appropriate responses using rich probabilistic representations. The paper also examines the question of rigorous validation of complex human-robot interfaces by evaluating the proposed interface in the context of a standardized rehabilitation task domain.

A Computational Model of Graded Cueing: Robots Encouraging Behavior Change

“…Various robots have previously been used to shape child behavior [16,3,5]. Robins et al [16] found that four children with ASD imitated a doll-like robotic toy, often without any initial prompting; they attributed this to the robot's "simpler" physical appearance.…”

“…Duquette et al [3] found that a humanoid robot elicited more shared attention between two children with ASD than a human mediator. Ferrari et al [5] presented the IROMEC robot as a social mediator in interactive play between a child with ASD and a parent, teacher or therapist. Our work extends previous work by Feil-Seifer and Matarić, which introduced the use of graded cueing applied to socially assistive robots interacting with children with ASD [4].…”

Design and Evaluation of a Flexible Interface for Spatial Navigation

“…To allow the user to make maximal use of these capabilities, it is important that the navigational targets be easy to select for users with a wide range of motion impairments. The smart wheelchair's onboard computer system is equipped with standard robotic software (mapping, localization, path planning and obstacle avoidance, see [11] for details) allowing it to reach the selected targets.…”

Toward Humanlike Task‐Based Dialogue Processing for Human Robot Interaction

“…For one, robotic systems that use natural language for robot instruction either do not have natural language fully integrated into the robotic architecture (for example, Michalowski et al [2007]) or are limited to simple instructions (for example, Firby [1989], Atrash et al [2009]). And many systems use rule-based grammars due to the difficulty of producing a well-trained grammar on what is invariably the small amount of data applicable to the domain.…”