Expert-in-the-Loop Multilateral Telerobotics for Haptics-Enabled Motor Function and Skills Development

摘要

Among medical robotics applications are Robotics-Assisted Mirror Rehabilitation Therapy (RAMRT) and Minimally-Invasive Surgical Training (RAMIST) that extensively rely on motor function development. Haptics-enabled expert-in-the-loop motor function development for such applications is made possible through multilateral telerobotic frameworks. While several studies have validated the benefits of haptic interaction with an expert in motor learning, contradictory results have also been reported. This emphasizes the need for further in-depth studies on the nature of human motor learning through haptic guidance and interaction. The objective of this study was to design and evaluate expert-in-the-loop multilateral telerobotic frameworks with stable and human-safe control loops that enable adaptive “hand-over-hand” haptic guidance for RAMRT and RAMIST.\nThe first prerequisite for such frameworks is active involvement of the patient or trainee, which requires the closed-loop system to remain stable in the presence of an adaptable time-varying dominance factor. To this end, a wave-variable controller is proposed in this study for conventional trilateral teleoperation systems such that system stability is guaranteed in the presence of a time-varying dominance factor and communication delay. Similar to other wave-variable approaches, the controller is initially developed for the Velocity-force Domain (VD) based on the well-known passivity assumption on the human arm in VD. The controller can be applied straightforwardly to the Position-force Domain (PD), eliminating position-error accumulation and position drift, provided that passivity of the human arm in PD is addressed. However, the latter has been ignored in the literature. Therefore, in this study, passivity of the human arm in PD is investigated using mathematical analysis, experimentation as well as user studies involving 12 participants and 48 trials. The results, in conjunction with the proposed wave-variables, can be used to guarantee closed-loop PD stability of the supervised trilateral teleoperation system in its classical format. The classic dual-user teleoperation architecture does not, however, fully satisfy the requirements for properly imparting motor function (skills) in RAMRT (RAMIST). Consequently, the next part of this study focuses on designing novel supervised trilateral frameworks for providing motor learning in RAMRT and RAMIST, each customized according to the requirements of the application.\nThe framework proposed for RAMRT includes the following features: a) therapist-in-the-loop mirror therapy; b) haptic feedback to the therapist from the patient side; c) assist-as-needed therapy realized through an adaptive Guidance Virtual Fixture (GVF); and d) real-time task-independent and patient-specific motor-function assessment. Closed-loop stability of the proposed framework is investigated using a combination of the Circle Criterion and the Small-Gain Theorem. The stability analysis addresses the instabilities caused by: a) communication delays between the therapist and the patient, facilitating haptics-enabled tele- or in-home rehabilitation; and b) the integration of the time-varying nonlinear GVF element into the delayed system. The platform is experimentally evaluated on a trilateral rehabilitation setup consisting of two Quanser rehabilitation robots and one Quanser HD2 robot.\nThe framework proposed for RAMIST includes the following features: a) haptics-enabled expert-in-the-loop surgical training; b) adaptive expertise-oriented training, realized through a Fuzzy Interface System, which actively engages the trainees while providing them with appropriate skills-oriented levels of training; and c) task-independent skills assessment. Closed-loop stability of the architecture is analyzed using the Circle Criterion in the presence and absence of haptic feedback of tool-tissue interactions. In addition to the time-varying elements of the system, the stability analy