Adaptive Haptic Forces in a Virtual Environment Improve Fine Motor Skill Training

Investigations of technological interventions to retrain motor skills using haptic control have included techniques that guide the patient (e.g., virtual fixtures) or challenge the patient (e.g., error amplification). Virtual fixtures are force fields or channels presented by a haptic device that prevent participants from deviating from a defined path based on expert performance. Error amplification increases the magnitude of errors when participants deviate from the desired path; that is, haptic forces are applied away from the desired path when errors occur.;Other training systems have incorporated adaptive aiding with haptic guidance. With adaptive aiding the intensity of virtual fixtures is modified (decreased) as operator performance improves. There is evidence that providing guidance when needed is more effective than a constant or fixed amount of assistance. Although haptic guidance has been found to enhance performance when engaged, it may not accelerate learning. To date, studies on learning effects of haptic guidance have not included adaptive aiding and haptic control using error amplification.;This research prototyped and evaluated a system that combined error amplification with adaptive aiding to determine the extent to which performance in a progressive error amplification condition transfers to an unassisted condition (i.e., a measure of skill learning). Data collection included two phases, including: (1) unassisted drawing with a haptic device; and (2) drawing with a form of haptic guidance. In both phases participants were trained to draw a series of letter-like designs (letters from a foreign alphabet) with the non-dominant hand. Phase 1 was used to determine learning rates for participants drawing the letters without guidance in order to inform adaptive aiding schedules for Phase 2. In Phase 2, each participant received one of four forms of haptic guidance, including adaptive virtual fixtures, static virtual fixtures, adaptive error amplification or static error amplification (following a between-subjects experiment design). As participant performance improved under adaptive conditions, haptic aiding was modified; virtual fixtures were reduced and error amplification was increased. Improvements were measured objectively with comparisons of motion trajectories to a template (accuracy) and task time (speed). An unassisted test was presented after multiple training sessions. Performance was compared among the four guidance conditions to determine how training with virtual fixtures compared to error amplification with and without adaptive aiding, once haptic guidance is no longer engaged.;Results of the study were mixed. Task accuracy when training with virtual fixtures was greater than the error amplification condition, but improved virtual fixture accuracy did not transfer to the test scenario. Test accuracy following error amplification training was greater. However, task speed improvements were higher for virtual fixtures than error amplification. The error amplification condition provided more constant feedback, which caused participants to more frequently evaluate their progress, especially at greater difficulty levels.;These findings are expected to advance the design of VR systems for fine motor skill training. Future writing tutors as well as training programs for occupational domains where fine motor skills are required could benefit. Low level performance and learning data from the experiment could be used as a basis for developing a motor capability classification algorithm that could quickly assess a person's level of fine motor skill. This algorithm could be used in healthcare for determining appropriate therapy regimens for people recovering from impairment or in industry for identifying training levels for jobs requiring fine motor control.