成果報告書詳細
管理番号20160000000579
タイトル*平成27年度中間年報 次世代ロボット中核技術開発 (革新的ロボット要素技術分野)ブレイン・マシン・インターフェース 脳波によるヒト型ロボット高速制御技術の実現可能性に関する検討
公開日2016/7/22
報告書年度2015 - 2015
委託先名国立研究開発法人産業技術総合研究所
プロジェクト番号P15009
部署名ロボット・AI部
和文要約
英文要約Title: An attempt of speed-up of an EEG-based BMI system to control the robot avatar.(FY2015-FY2016) FY2015 Annual Report

 AIST has been developing a brainwave-based brain-machine interface (BMI), Neurocommunicator as a communication aid for patients with severe motor deficits. In order to enable the patients to express their feelings and requests to their family and caregivers, the system focuses on the event-related potential (ERP) that reflects an increase in top-down attention. The user of the system wears a compact headgear and looks at eight pictograms in a matrix on a PC display, which are the candidates of messages. Flashes of a word with 4 Japanese characters (Ko-Re-Ka-Na) elicited a strong ERP when the target pictogram is flashed, and the response is automatically detected by the real-time pattern recognition technique.
 In this study, we are investigating a possibility to apply the Neurocommunicator to the robot control. We first focused on the introduction of the robot avatar to express a message of the user, which is originally expressed by the CG avatar. The main goal of this study at the first budget year is speed-up of the neural decoding technique for a quick control of the robot. So far, we have already achieved good results (94% accuracy at 3.74 sec on average of 10 subjects) by the high-speed presentation of the stimuli as well as by the special algorithm (virtual decision function: VDF), which is close to the desired goal (95% accuracy at 3 sec).
 Second, we also plan to add a function to navigate the robot avatar to the system, and developed a new interface with a small cursor to elicit the ERP. We will test this interface for several subjects.
 Third, we started using the robot avatar (mainly by the video library) to the clinical study. For a few patients have enjoyed the robot control by the Neurocommunicator.
ダウンロード成果報告書データベース(ユーザ登録必須)から、ダウンロードしてください。

▲トップに戻る