Unified Model and Robotic Implementation of Bio-Inspired Walking and Running

该奖项的研究目标是开发和实施一个生物力学相关的，统一的腿动力学理论，跨越步行和跑步，并在一个新的双足机器人ATRIAS上展示这一理论。迄今为止，还不存在动态行为可以接近人类行走和跑步的性能（包括这两种步态之间的过渡）的腿式机器。这项研究将导致步态和步态过渡与人类一样的腿部动力学的原则模型，在广义模型操纵循环混合动力系统，以实现不同的目标行为，并在验证和演示这一新的科学认识与双足机器人。研究进展从分析简化的步态模型，捕捉人体运动的基本动力学集成这些模块在一个核心的人体步态动力学模型。目标行为操纵将接近使用的影响参数的庞加莱映射的形状在这些循环动力系统。对机器人的实验验证将展示和完善理论进展。可提供的内容包括建模和仿真软件、机器人硬件的实验验证和演示、研究成果文档以及工程专业学生的教育和研究经验。这项研究的结果将提供机会，为动力腿系统创建控制算法，使功能多样的行为类似于动物和人类。示例包括假腿和外骨骼，其在行走、跑步以及这些步态之间的过渡期间为人类用户提供类似人类的腿部动力学。其他示例包括实现鲁棒且高效的动态转向行为的腿式机器人。研究结果将予以传播，以鼓励商业应用和进一步开发。机器人实现结果的模拟将在网上发布，以参与动力腿控制算法开发的健康竞争。研究生和本科生将通过课堂整合，研究参与和“指导树”受益，支持人员将为本科机器人俱乐部的学生提供建议和指导。

The research objectives of this award are to develop and implement a biomechanically relevant, unified theory of legged dynamics that spans walking and running, and to demonstrate this theory on a new bipedal robot, ATRIAS. To date, a legged machine whose dynamic behavior can approach the performance of human walking and running, including transitions between these two gaits, does not exist. The research will result in principled models of gait and gait transitions with human-like leg dynamics, in generalized models for manipulating cyclic hybrid dynamic systems to achieve different goal behaviors, and in the verification and demonstration of this new scientific understanding with a bipedal robot. The research progresses from analyzing simplified gait models that capture the essential dynamics of human locomotion to integrating these modules in a core dynamics model of human gait. Goal behavior manipulations will be approached using the influence of parameters on the shape of the Poincare map in these cyclic dynamic systems. Experimental verification on the robot will demonstrate and refine the theoretical progress. Deliverables include modeling and simulation software, experimental verification and demonstration on robot hardware, documentation of research results, and engineering student education and research experience. The results of this research will provide the opportunity to create control algorithms for powered legged systems that enable functionally versatile behaviors similar to animals and humans. Examples include prosthetic legs and exoskeletons that provide human users with human-like leg dynamics during walking, running, and the transitions between these gaits. Other examples include legged robots that achieve robust and efficient dynamic turning behaviors. The results will be disseminated to encourage commercial applications and further development. A simulation of the robot implementation of the results will be posted online to engage healthy competition in the development of control algorithms for powered legs. Graduate and undergraduate students will benefit through classroom integration, research involvement, and a "tree of mentoring" where the supported personnel will advise and mentor undergraduate robotics club students.