Towards multi-tasking teleoperated robot teams

点击翻译按钮获取中文摘要

Teleoperated robots are robots controlled by human users from afar. They are essential for tasks that demand swift responses to unpredictable emergencies in unknown and dynamic environments in which humans cannot be deployed because those environments are inaccessible or dangerous. The needs for teleoperation are accelerating in the clean energy (maintenance of off-shore wind and underwater installations), industry 4.0, robotics, and transportation industries. The global market is predicted to reach $76.5bn US by 2026 and to grow at a rate of over 20% year-over-year until 2050. This robust industry demand is not only for more pervasive but also for increasingly capable teleoperation. In response to the industry need for more skillful teleoperation, this proposal offers to endow teleoperated robot teams with multi-tasking capabilities. The overarching goal is to empower remote robot teams to execute several, possibly conflicting, teleoperation commands at once. Compared to the current teleoperation paradigm - one or multiple robots carrying out a single task under the guidance of one or several human operators - multi-tasking teleoperated robot teams will have increased flexibility, agility and resilience. They will use their resources more fully and, thus, more economically. They will also find application in important practical scenarios in which the current teleoperation paradigm is insufficient. For example, when fighting a wide-spread and rapidly advancing fire, each robot of a remote team of robots may face a different emergency at the same time as the other robots face their emergencies. Thus, multiple users must teleoperate the robot team to survey the environment and to respond to all emergencies simultaneously. In the short-term, the research program has three objectives - to develop analysis and control design frameworks: 1) for multiuser tele-navigation of robot networks with switching spanning trees; 2) for multiuser tele-navigation of resilient robot networks; and 3) for multiuser tele-manipulation of robot networks. Each short-term objective entails formulation of its control problem and an associated distributed decision-making and resilience dynamics, discovery of bounded potentials to maintain the connectivity of the robot network, passive coupling of the decision and of the resilience dynamics to the dynamics of the robot network, and appropriate design of the coupling of the robot network to the multiple users to render the overall teleoperator system passive with respect to the multidimensional power ports that connect it to the users and the environment. The innovations - in the designs of the decision and resilience dynamics and of the couplings between various subsystems - will advance both teleoperation control and, more generally, the control of multirobot systems.