Haptic-Enabled Collaborative Virtual Environments for Skills Training

Abstract

<p> Many manual tasks such as those in surgical applications require a high degree of motor skills that can only be gained through extensive training. This thesis is concerned with the design and control of collaborative training virtual environments with haptic feedback for skills training. The term "collaborative training" refers to a scheme in which the trainee and the trainer operate in a shared virtual environment. They collaboratively carry out the intended tasks using a shared "virtual tool". In order to enhance the trainee's motor skills, the conventional visual feedback will be augmented by force feedback providing the feel of the task environment as well as active guidance by the expert trainer.</p> <p> First, a set of psychophysics experiments are designed to investigate the usefulness of haptic-enabled collaborative virtual environments for motor skills training. Eighteen volunteers randomly divided between two training and control groups have participated in the experiments. The training group would undergo a number of collaborative training sessions with active help from the trainer whereas the control group would try the task on their own to achieve a set of stated goals. Each of the experiments is designed with specific performance objectives in mind, including trajectory tracking and task completion time. The results of the psychophysics experiments confirm that, when visual feedback is partially impaired, haptic-enabled collaborative training improves learning of a trajectory tracking task. In all the experimental scenarios tested, the results showed improvements in temporal response after receiving training.</p> <p> The second part of the thesis is devoted to the development of a general control framework for the coordination of the users in haptic-enabled collaborative virtual environments. The haptic interface control design is separated from the virtual environment simulation in order to provide more versatility in control strategies for both impedance and admittance-type virtual environments. Adaptive nonlinear controllers are proposed that establish desired linear-time-invariant and/or nonlinear static mappings amongst the users and the virtual task environment positions and forces. These controllers account for the nonlinear model of haptic devices and can handle uncertainties in the haptic devices, the users, and the virtual environment dynamics. First, the tracking behavior of the system is shown via a Lyapunov analysis. Then using a priori known bounds on user and environment parameters, the robust stability of the system is analyzed by employing the Nyquist envelops of interval plants and an off-axis circle criterion. The robust stability analysis provides bounds on the parameter of the linear and nonlinear mappings within which the stability of the system is guaranteed, for all possible system parameters with their a priori given bounds. Experiments carried out with two similar Quanser twin-pantograph haptic devices confirm the effectiveness of the proposed controllers in achieving the performance and stability objectives.</p>