Human-state based active physical human-robot interaction

点击翻译按钮获取中文摘要

Collaborative manipulators and assistive robotic devices have increased in sales and adoption in the last decade, indicating a shift towards applications involving humans working side by side with robots, both in industrial applications and domestic environments. Physical interactions with robotic devices will become inevitable due to rapid advances in technology. Research on robot control has heavily focused on guaranteeing physical safety, but often ignores the psychological safety and mental comfort of the human user. This program intends to explore the perceptions and mental comfort of human users in the control of robots, enabling a truly safe human-robot interaction, from both the mental and physical points of view. The proposed research program consists of the following objectives: 1. To investigate the perceptions related to physical interactions with robots and to identify the relevant factors (e.g., acceptance, mental load, and exchanged forces); 2. To build a human-state model that is comprehensive of both mental and physical states based on real-time measurable data, considering also improvements in interaction history using machine learning methods; 3. To develop an adaptive and predictive controller for robots that uses the human-state model to interact with users, so that the control structure will be hierarchical and will consider at the same time the state of the human, the environment, and task completion; and 4. To validate the outcomes on actual robot platforms, starting with the collaborative manipulator, which is a state-of-the-art robot being produced worldwide and in Canada on a large scale, with Canadian companies such as Kinova and companies with production sites in Canada such as ABB, KUKA. This research program is highly interdisciplinary as it includes aspects of social robotics, robot control, human modeling, human factors, and machine learning. The facilities at the University of Waterloo (UW), including the RoboHub and the Motion Capture Lab (MCL) will permit this program to reach its defined objectives, and contribute to the training of highly skilled professionals with cutting--edge technology. Our students will become specialized in human-robot interaction, robot control, modeling, and machine learning, which play essential roles in the robotics industry and robotics applications, such as industrial warehouse organization, assembly lines, intelligent homes, care industry, aiming to improve the quality of life of Canadians. The trained personnel from this program will contribute to advancements in robotics in both academia and industry, which will establish the role of Canada as world leader in the rising field of robotic technologies.