基于身体归属脑机制的人机共融轮椅机器人行为意图实时准确认知与人机协同研究

Aiming at the novel study challenges in the core technologies for the new-generation cooperative robots to serve the aged and disabled people, and focusing on the core scientific issue of accurate cognition and understanding of human behavioral intentions and human-machine coordination as well as natural interaction, this project will carry out innovative interdisciplinary research on the cognitive theory and method of enhanced EEG signals through interaction of body-ownership mechanism based on perceptual induction of vision and tactility recently revealed by brain science and active motor imagery in human motor cortex, in order to achieve accurate cognition and understanding of human behavioral intentions in the cooperative robot BCI systems; investigating the methods for BCI fusion with machine vision and online autonomous learning, to impement the active multimode perception and natural interaction of human-machine-environment and achieve the intelligent coordination of human-machine cooperation in the complex dynamic environment; accomplishing the reliability and security of the human-machine cooperative operations by further inducing the error correction mechanism based on the EEG's Error-Related Negativity signal; and finally testing and improving the prototype of the human-machine cooperative wheelchair robot by involving both healthy and disabled participants from the perspective of application verification. Through the research, this project will provide novel basic theories and feasible key technologies for the research and application of new-generation service cooperative-robots.

本项目面向助老助残服务新一代共融机器人核心技术研究挑战，围绕人体行为意图准确认知理解与人机协同及自然交互这个核心科学问题进行跨学科交叉创新研究。通过研究脑科学最新揭示的基于视触觉感知诱发的身体归属机制与主动性运动想象在大脑运动皮层交互增强的脑电信号认知理论及方法，以实现共融机器人BCI系统对操作者行为意图的准确认知理解；通过研究BCI与机器视觉感知融合和机器人在线自主学习方法，以实现人-机-环境的多模态主动感知与自然交互、从而达成在复杂动态环境中人机共融的智能人机协同；并通过进一步引入基于错误相关电位的纠错机制，以达成人机共融作业中的安全与可靠；最后，从应用验证角度，分别使用健康和残障受试者对人机共融轮椅机器人原型系统进行测试和改进。通过本项目的研究，将能为新一代服务型共融机器人的研究及应用提供国际前沿创新的基础理论与切实可行的关键技术。

当前机器人研究朝着自主化智能化和共融化发展，包含机器人运动学习、机器人与环境的共融交互以及与人类间的协同控制等关键问题，这也要求共融机器人能够认知环境并理解人类的意图。本课题组围绕项目计划书，开展了运用多模式脑机接口（BCI）以实现对操作者行为意图认知理解研究、运动皮层脑神经机制、以及人机动作行为等类脑运动编码学习与运动控制等跨学科交叉研究，通过这些研究成果目前已发表了15篇SCI论文和5篇机器人国际会议论文。并且，课题组还构建了搭载有视觉、语音以及脑机接口等多传感的人机共融轮椅机器人实验平台，以便从应用角度验证和完善理论研究的实际可行性。