Masafumi Uchihara

Human-robot collaboration in precise positioning of a three-dimensional object

This paper deals with fundamental issues of human-robot cooperation in precise positioning of a flat object on a target. Based on the analysis of human-human interaction, two cooperation schemes are introduced. Several algorithms implementing these schemes are developed. A general theoretical framework for human/robot cooperation has been developed to represent these algorithms. The evaluation of the algorithms was carried out using our in-house made robot prototype and experiments by human subjects has demonstrated the effectiveness of our schemes. The main problem was the regulation of the robot-human interaction. Since the robot has no range sensors, it has to rely on the force and displacement information resulting from the interaction with human to understand human intention. The way the robot interprets these signal is crucial for smooth interaction. To be able to carry out a concrete task a simplification was made, in which robot and human do not directly hold the object but a frame to which the object and various sensors are attached.

Human-Robot Cooperation in Precise Positioning of a Flat Object

Abstract This paper deals with fundamental issues of human-robot cooperation in precise positioning of a flat object on a target. Based on the analysis of human-human interaction, two cooperation schemes are defined. An algorithm imitating one of these schemes is presented. A general mathematical framework for human/robot cooperation has been developed, based on which several algorithms are proposed. The evaluation of the algorithms was carried out using our in-house made robot prototype using a number of subjects has demonstrated the effectiveness of our ideas. The main problem was the regulation of the robot human interaction. Since the robot has no range sensors, it has to rely on the force and displacement information resulting from the interaction with human. The way the robot interprets these signal is crucial for smooth interaction. To be able to carry out a concrete task a simplification was made, in which robot and human do not directly hold the object but a frame to which the object and various sensors are attached. Based on our research results, we succeeded in installing a commercial platform.

A Feedback/Feedforward Accommodation of Load Change and its Application to Control of Reformer for Fuel Cell Vehicle

Abstract Output regulation for nonlinear plants with large load variation is considered assuming that a nonlinear model of the plant for simulation is available. A linear model of the plant for a fixed load is obtained via 4SID method which naturally leads to a feedback/feedforward control architecture. The control configuration is applied to control of reformer for fuel cell vehicle, where H∞ control is applied for feedback. It is shown both theoretically and experimentary that the overall feedback/feedforward configuration achieves a high performance of regulation under heavy load change.