Passivity-based bilateral human-swarm-interactions for cooperative robotic networks and human passivity analysis

Abstract

This paper investigates bilateral human-swarminteractions wherein the objective is to guarantee human operator enabled synchronization of positions/velocities of an ensemble of kinematic robots to desired reference inputs. We first present a feedback loop configuration, where every robot implements a cooperative controller and the human visually feedbacks the average positions/velocities of the accessible robots depending on the selected (position or velocity) control modes. Asymptotic synchronization is demonstrated by assuming passivity of an appropriately defined human operator decision process. The aforementioned passivity assumption and learning ability of the operator are studied through experiments on a human-in-the-loop simulator. It is observed that learning has a positive effect on passivity, but passivity of the decision process may nevertheless be violated in the high frequency domain depending on the network connection. Hence, passivation scheme is presented for the operator's decision process and it is demonstrated for three different interconnection structures among the kinematic robots and operator.

Robot team teleoperation for cooperative manipulation using wearable haptics

“…The interaction is achieved via a haptic device. Teleoperation of robot teams in which the human operator is not coupled to a haptic device is investigated in [4] and [5] (hand motion tracked with cameras) and in [6] (commands provided via a tablet). However, the feedback provided to the human in this case is visual, which is insufficient when the robots interact with the environment.…”

Control sharing in human-robot team interaction

“…However, the human is a point of failure of such a system [11]. One solution to this problem is the interaction with a subset of robots in the team [54].…”

“…In classical teleoperation literature, control design commonly relies on the assumption that the trained human behaves passive [56,103]. The authors of [54] use black-box methods to identify human decision-making behavior in the active role of commanding a robot swarm. The frequency analysis of the obtained linear time-invariant system, however, reveals that the human decisionmaking process violates the passivity condition in the high-frequency range.…”

“…The choice can be made between global intervention (steering the centroid of the formation to the goal) or local intervention (steering a single UAV). In [54] the human switches manually between two controllers: the control of the robot team position and the control of the robot team velocity, and provides the input commands to the chosen controller.…”

“…In the active role, the human provides continuous commands, for example motion and/or force commands. This type of commands is typically conveyed through a motion measuring device such as a touch screen [54], vision-based system [102], [2], or a haptic device [64,81].…”