In shared control teleoperation, the robot assists the user in accomplishing the desired task, making teleoperation easier and more seamless. Rather than simply executing the user's input, which is hindered by the inadequacies of the interface, the robot attempts to predict the user's intent, and assists in accomplishing it. In this work, we are interested in the scientific underpinnings of assistance: we propose an intuitive formalism that captures assistance as policy blending, illustrate how some of the existing techniques for shared control instantiate it, and provide a principled analysis of its main components: prediction of user intent and its arbitration with the user input. We define the prediction problem, with foundations in inverse reinforcement learning, discuss simplifying assumptions that make it tractable, and test these on data from users teleoperating a robotic manipulator. We define the arbitration problem from a control-theoretic perspective, and turn our attention to what users consider good arbitration. We conduct a user study that analyzes the effect of different factors on the performance of assistance, indicating that arbitration should be contextual: it should depend on the robot's confidence in itself and in the user, and even the particulars of the user. Based on the study, we discuss challenges and opportunities that a robot sharing the control with the user might face: adaptation to the context and the user, legibility of behavior, and the closed loop between prediction and user behavior.

/pdf/a-policy-blending-formalism-for-shared-control-496mi2akqp.pdf

A policy-blending formalism for shared control

A medical system that allows a mentor to teach a pupil how to use a robotically controlled medical instrument. The system may include a first handle that can be controlled by a mentor to move the medical instrument. The system may further have a second handle that can be moved by a pupil to control the same instrument. Deviations between movement of the handles by the mentor and the pupil can be provided as force feedback to the pupil and mentor handles. The force feedback pushes the pupil's hand to correspond with the mentor's handle movement. The force feedback will also push the mentor's hand to provide information to the mentor on pupil's movements. The mentor is thus able to guide the pupil's hands through force feedback of the pupil handles to teach the pupil how to use the system.

Minimally invasive surgical training using robotics and telecollaboration

A multi-user medical robotic system for collaboration or training in minimally invasive surgical procedures includes first and second master input devices, a first slave robotic mechanism, and at least one processor configured to generate a first slave command for the first slave robotic mechanism by switchably using one or both of a first command indicative of manipulation of the first master input device by a first user and a second command indicative of manipulation of the second master input device by a second user. To facilitate the collaboration or training, both first and second users communicate with each other through an audio system and see the minimally invasive surgery site on first and second displays respectively viewable by the first and second users.

Multi-user medical robotic system for collaboration or training in minimally invasive surgical procedures

While motor interaction between a robot and a human, or between humans, has important implications for society as well as promising applications, little research has been devoted to its investigation. In particular, it is important to understand the different ways two agents can interact and generate suitable interactive behaviors. Towards this end, this paper introduces a framework for the description and implementation of interactive behaviors of two agents performing a joint motor task. A taxonomy of interactive behaviors is introduced, which can classify tasks and cost functions that represent the way each agent interacts. The role of an agent interacting during a motor task can be directly explained from the cost function this agent is minimizing and the task constraints. The novel framework is used to interpret and classify previous works on human-robot motor interaction. Its implementation power is demonstrated by simulating representative interactions of two humans. It also enables us to interpret and explain the role distribution and switching between roles when performing joint motor tasks.

/pdf/a-framework-to-describe-analyze-and-generate-interactive-1yen5w5sxt.pdf

A Framework to Describe, Analyze and Generate Interactive Motor Behaviors

This chapter sets about to provide the background and orientation needed to set a novice designer on his or her way to bringing haptics successfully into an interactive product. To define appropria...

Haptic Interaction Design for Everyday Interfaces

In this brief, we propose a "shared-control" architecture for design of a haptic interface for hands-on training in minimally invasive surgery. The interface is designed to allow experienced surgeons mentor trainee surgeons through shared control of a surgical robot. During shared control, the interface provides feedback forces to both surgeons, proportional to the difference of their actions and reversely proportional to the control authority shared between them. The control authority shared between the surgeons is chosen based on their relative level of surgical skills and experience and determines the extent to which the motion of the surgical robot depends on their individual commands. We use the H/sub /spl infin// approach to design the haptic interface and present experimental results to demonstrate its suitability for the intended application of minimally invasive surgical training.

M. Ghodoussi

Papers

A shared-control approach to haptic interface design for minimally invasive telesurgical training