Closed-loop control is important for amputees to manipulate myoelectric prostheses intuitively and dexterously. Tactile feedback can help amputees improve myoelectric control performance for grasping objects. To investigate the effects of different tactile feedback, we performed experiments on six amputees and six able-bodied subjects via electrotactile stimulation. Using a virtual environment, six kinds of objects with different weights and stiffnesses were used for grasping tasks. Five feedback conditions (no feedback, pressure feedback, slip feedback, pressure + slip feedback, and vision feedback) were considered. Nine evaluation indexes and three control objectives (rapidity, economy, and stability) were proposed. Under the five feedback conditions, our study investigated four issues: 1) three types of grasping-related failures; 2) four types of grasping-related time measures; 3) average grasping force; 4) standard deviation of the grasping force. Results indicate that: 1) slip feedback is better than pressure feedback; 2) pressure + slip feedback can improve grasping rapidity; 3) slip feedback significantly contributes to grasping economy and stability; and 4) pressure + slip feedback can perform as well as vision feedback.

Virtual environments (VEs) with haptic feedback not only provide a safe and versatile practice medium for many manual control tasks, but also promise to improve the efficacy of manual skill training with the addition of haptic guidance. However, haptic guidance schemes such as shared control may be detrimental since such schemes actively interfere with the coupled system dynamics, thereby causing participants to experience task dynamics that are altered from those of the real task. Therefore, this paper proposes a progressive approach to guidance for training in virtual environments. This progressive guidance scheme adjusts its control gains based on participant performance, aiming to expose the performer to an appropriate amount of haptic guidance throughout training. Long term training experiments were conducted for an under-actuated target-hitting manual control task. The experimental results compare the efficacy of the novel progressive haptic guidance to two common fixed-gain haptic guidance schemes and virtual practice. The results from a month-long training experiment indicate that the proposed progressive shared control scheme reduces guidance interference as compared to fixed-gain guidance schemes thus increasing training efficacy.

This paper presents the design and implementation of a novel progressive haptic guidance scheme and a similar visual guidance scheme for acquisition of a dynamic motor skill. The paper compares the schemes' performance to each other and to practice alone without any form of guidance. The target-hitting task is represented in a visual and haptic virtual environment and implemented in a training protocol that lasts eleven sessions over a two-month period. The progressive guidance controller employs as inputs two expertise-based performance measures, trajectory error and input frequency. The analysis of the experimental results demonstrates that while guidance is active, haptic guidance outperforms both visual guidance and practice alone (no guidance) until late in the protocol when all three groups saturate at the same level of performance. The results fail to show significant differences in training outcomes because the performance of all participants saturates toward the end of the protocol. The key implication of the experimental findings is that visual and haptic guidance presented in a progressive manner have no detrimental effects on performance. Our results confirm that haptic guidance, based on skill component measures, is effective early in the training protocol when participants are only beginning to understand the components of the task but should be progressively removed to avoid possible negative dependence on the guidance.