基于多模态意图识别和变参数导纳的踝康复机器人柔顺交互训练研究

Aiming at the problem of insufficient compliance and single rehabilitation training of existing ankle rehabilitation robot, we first systematically study the quantification research and compensation of the additional force/moment generated in the process of the active rehabilitation training. Then, we study the multi-coupling dynamics model, joint space mapping and position resolution of human-machine system, ankle’s motion intention recognition based on multi-modal information fusion, and compliant human-computer interaction control strategy with adaptive adjustment of admittance parameters in turn. On this basis, we aim to design and develop the multi-mode rehabilitation training technology with active rehabilitation training function as the core, combined with active and passive compliant rehabilitation training for the training of range of motion and the muscle strength of the ankle joint. And all the above researches will be realized in the designed 2-UPS/ RRR parallel ankle rehabilitation robot with ankle rehabilitation training experiments. Through the research of this project, we will reveal the influence of the lower limb’s posture change on the additional force/moment, establish the accurate recognition model of the ankle’s motion intention, and propose the compliant interactive control and multi-mode rehabilitation training method for ankle rehabilitation robots, which will provide the theoretical basis for the development of scientific rehabilitation training plans. The implementation of the project can not only solve the common problems in the existing ankle rehabilitation robots, but also help to facilitate the application of parallel ankle rehabilitation robots to the clinical practice, which has important theoretical significance and application value.

针对现有踝康复机器人柔顺性不足、康复训练方式单一的问题，本项目从主动康复训练过程中附加力/矩量化研究与补偿入手，拟通过对“人机闭链的多耦合动力学模型、关节空间映射及位置解析”、“基于多模态信息融合的踝运动意图识别”、“导纳参数自适应调整的柔顺人机交互控制策略”等内容的研究，设计和研发以主动康复训练功能为核心，主/被动柔顺康复训练相结合的、针对踝关节活动度和肌力训练的多模式康复训练技术，并在课题组研发的2-UPS/RRR并联踝康复机器人上进行功能实现与康复训练的试验研究。通过本项目的研究，揭示下肢姿态变化对附加力/矩的影响规律，建立踝运动意图的准确识别模型，提出针对踝康复机器人的柔顺交互控制及多模式康复训练方法，为制定科学的康复训练计划提供依据。该项目的实施不仅可以解决现有踝康复机器人存在的共性问题，而且有助于推动并联踝康复机器人走向临床应用，具有重要的理论意义和应用价值。

本项目从附加力/矩量化研究与补偿入手，通过对“人机闭链的多耦合动力学模型、关节空间映射及位置解析”、“基于力位传感的踝运动意图识别和导纳参数自适应调整的柔顺人机交互控制策略”、“踝康复机器人多模式康复训练与康复训练舒适度的数值化评估”等内容的研究，设计和研发以主动康复训练功能为核心，主/被动柔顺康复训练相结合的踝康复训练技术，并在2-UPS/RRR踝康复机器人上进行功能实现与康复训练的试验研究。具体的成果包括：（1）结合姿态传感器，揭示下肢姿态变化对附加力/矩的影响规律，完成了并联踝康复机器人的关节空间映射及位置解析，建立了该机构的运动学标定算法，并实现了被动康复训练轨迹的规划设计；（2）结合附加力/矩的量化研究，以及人机交互力矩的提取，探究基于力位信息的运动意图识别算法，并搭建导纳控制系统框架，实现了基于导纳模型的柔顺交互，在此基础上，提出了被动柔顺训练、等张训练和主动训练等多种柔顺交互训练模式，并探究了基于力位的自适应导纳参数调整算法；（3）探究了基于sEMG的肌肉激活评估，提出了一种基于样本熵估计肌肉激活状态的方法，将样本熵引入传统的肌肉激活模型，并对其进行修正，实现对机器人辅助踝康复训练效果的评估。