CPS: TTP Option: Synergy: Learning and Adaption in Pediatric Robotics

CPS：TTP 选项：协同：儿科机器人的学习和适应

• 批准号: 1545106
• 负责人: Homayoon Kazerooni
• 金额: $ 120万
• 依托单位: University of California-Berkeley
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2015
• 资助国家: 美国
• 起止时间: 2015-10-01 至 2019-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1545106&HistoricalAwards=false
• 关键词: CPS; TTP; Option; Synergy; Learning

Children affected by neurological conditions (e.g., Cerebral Palsy, Muscular Atrophy, Spina Bifida and Severe head trauma) often develop significant disabilities including impaired motor control. In many cases, walking becomes a non-functional and exhausting skill that demands the use of the aids or the substitution of function, such as wheelchair. This usually cause these children not to acquire locomotion skills, and consequently to lose their independence. However, it is well understood that bipedal locomotion, an essential human characteristic, ensures the best physiological motor pattern acquisition. For this reason, in children with neurological and neuromuscular diseases, independent walking is a significant rehabilitation goal that must be pursued in a specific temporal window due to the plasticity of central nervous system. In other words, children with neurological conditions have a small window of time to acquire locomotion skills through assisted walking rehearsals. The objective of this research work is to create and experimentally validate a set of technologies that form the framework for the development of adaptive, self-balancing, and modular exoskeleton robotics systems for children with neurological disorders. It is our belief that the exoskeleton (and its associated infrastructure) resulting from this research will offer an effective tool to promote locomotion skill acquisition, and in general health, during a critical period in the early life of children with neurological conditions. This research proposal develops a data-driven human-machine modeling specific to physiological conditions. This creates regression models that predict the user behavior without explicit modeling the complex human musculoskeletal dynamics and motor control mechanism. Additionally this research project formulates a safe adaptive control problem as a model predictive control (MPC) problem. In this method, an optimal input sequence is computed by solving a constrained finite-time optimal control problem where exoskeleton intrusion (input from exoskeleton) is minimized to maximize the user's intent to promote learning. This project further develops a novel approach for stabilizing and preventing fall of the exoskeleton and the child as a whole. This method allows a child wearing an exoskeleton to learn locomotion skills described above with less likelihood of falls. This research project furthermore evaluates the developed technologies in terms of efficiency and efficacy and creates a novel fun game using exoskeleton for children to promote locomotion skills.

受神经系统疾病（如脑瘫、肌肉萎缩、脊柱裂和严重头部创伤）影响的儿童通常会出现严重残疾，包括运动控制受损。在许多情况下，行走成为一种非功能性和累人的技能，需要使用辅助工具或替代功能，如轮椅。这通常会导致这些孩子无法获得运动技能，从而失去独立性。然而，众所周知，两足运动作为人类的基本特征，确保了最佳的生理运动模式获取。因此，在患有神经和神经肌肉疾病的儿童中，由于中枢神经系统的可塑性，独立行走是一个重要的康复目标，必须在特定的时间窗口内追求。换句话说，患有神经系统疾病的儿童只有一小段时间通过辅助行走排练来获得运动技能。这项研究工作的目的是创建并实验验证一套技术，这些技术构成了为患有神经系统疾病的儿童开发自适应、自平衡和模块化外骨骼机器人系统的框架。我们相信，这项研究产生的外骨骼（及其相关基础设施）将为促进运动技能的习得提供有效的工具，并在患有神经系统疾病的儿童早期生命的关键时期促进整体健康。本研究计划开发一个数据驱动的特定生理条件的人机建模。这创建了回归模型，预测用户行为，而没有明确建模复杂的人体肌肉骨骼动力学和运动控制机制。此外，本研究还将安全自适应控制问题作为模型预测控制（MPC）问题进行了阐述。该方法通过求解约束有限时间最优控制问题来计算最优输入序列，最小化外骨骼入侵（来自外骨骼的输入）以最大化用户促进学习的意图。该项目进一步发展了一种稳定和防止外骨骼和儿童整体坠落的新方法。这种方法允许佩戴外骨骼的儿童学习上述运动技能，减少跌倒的可能性。本研究项目进一步评估了已开发的技术的效率和功效，并为儿童创造了一个利用外骨骼促进运动技能的新颖有趣的游戏。