Annu Int Conf IEEE Eng Med Biol Soc. 2021 Nov;2021:6465-6469. doi: 10.1109/EMBC46164.2021.9629622.
Multiarticulate bionic hands are now capable of recreating the endogenous movements and grip patterns of the human hand, yet amputees continue to be dissatisfied with existing control strategies. One approach towards more dexterous and intuitive control is to create a semi-autonomous bionic hand that can synergistically aid a human with complex tasks. To that end, we have developed a bionic hand that can automatically detect and grasp nearby objects with minimal force using multi-modal fingertip sensors. We evaluated performance using a fragile-object task in which participants must move an object over a barrier without applying pressure above specified thresholds. Participants completed the task under three conditions: 1) with their native hand, 2) with the bionic hand using surface electromyography control, and 3) using the semi-autonomous bionic hand. We show that the semi-autonomous hand is extremely capable of completing this dexterous task and significantly outperforms a more traditional surface-electromyography controller. Furthermore, we show that the semi-autonomous bionic hand significantly increased users' grip precision and reduced users' perceived task workload. This work constitutes an important step towards more dexterous and intuitive bionic hands and serves as a foundation for future work on shared human-machine control for intelligent bionic systems.
多关节仿生手现在能够再现人手的内源性运动和握持模式，但截肢者仍然对现有的控制策略不满意。一种更灵活、更直观的控制方法是创建一种半自主仿生手，它可以与人类协同完成复杂任务。为此，我们开发了一种仿生手，它可以使用多模态指尖传感器自动检测和抓取附近的物体，只需施加最小的力。我们使用易碎物体任务来评估性能，参与者必须在不超过指定阈值的情况下将物体移动过障碍物。参与者在三种情况下完成任务：1）使用原生手，2）使用表面肌电图控制的仿生手，3）使用半自主仿生手。我们表明，半自主手非常能够完成这项灵巧的任务，并且明显优于更传统的表面肌电图控制器。此外，我们表明，半自主仿生手显著提高了用户的握持精度，降低了用户感知到的任务工作量。这项工作是朝着更灵活、更直观的仿生手迈出的重要一步，也是未来智能仿生系统人机共享控制研究的基础。
