Collaborative Research: Intelligent and Agile Robotic Legged Locomotion in Complex Environments: From Planning to Safety and Robust Control

该项目将通过提高机器人系统的行走能力，促进科学进步，国家繁荣和福利。虽然有腿机器人在机械上变得越来越复杂，但在现实世界中，它们的敏捷性、灵巧性和动态稳定性与生物机器人（如动物和人类）相比仍有很大差距。该项目的总体目标是创建一个正式的基础，以实现智能和快速的运动规划、安全和鲁棒控制，以实现自主腿式机器在现实世界中的众多敏捷行为。这项研究将通过控制理论、优化和智能的无缝集成，使越来越复杂的机器人无处不在，从而产生广泛的社会影响。这些机器人将能够在极端危险的情况下，如工业事故和自然灾害，有效地协助或作为人类的远程延伸。综合教育计划将通过设计新的课程产生广泛的影响。针对K-12学生、教师和代表性不足的少数民族的stem拓展计划，以及本科生参与研究。该项目将通过在控制和分析方面的关键创新，在很大程度上未被探索的智能和健壮的腿部运动领域推进知识。基于混合系统理论、优化和感知的智能和快速运动规划算法，将为复杂环境中腿部运动的动态模型创建。该项目将解决基于非线性控制和矩阵不等式的鲁棒最优控制器的综合，用于敏捷和灵巧的运动。它还将创建基于集合不变性和二次规划的安全关键控制算法，用于在未知环境中确保避障的鲁棒控制器的实际实现。为了弥合理论与实现之间的差距，本研究将通过在弗吉尼亚理工学院和加州理工学院的研究人员实验室中对两台先进的四足和两足机器人进行实验，将理论创新转化为实践。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This project will advance the progress of science, national prosperity and welfare, by enhancing walking capabilities of robotic systems. While legged robots are becoming more mechanically sophisticated, their agility, dexterity, and dynamic stability are still far from capabilities of their biological counterparts (such as animals and humans) in real world situations. The overarching goal of this project is to create a formal foundation to enable intelligent and fast motion planning, safety, and robust control for a multitude of agile behaviors of autonomous legged machines in real-world situations. This research will have broad societal impact through seamless integration of control theory, optimization, and intelligence to enable ubiquitous use of increasingly sophisticated robots. Such robots will be able to effectively assist, or serve as remote extensions of, humans in situations of extreme danger, such as industrial accidents and natural disasters. The integrated educational plan will have broad impact by designing new courses. STEM-based outreach initiative for K-12 students, teachers, and under-represented minorities, and engagement of undergraduate students in research.The project will advance knowledge in the largely unexplored field of intelligent and robust legged locomotion through key innovations in controls and analysis. Intelligent and fast motion planning algorithms, based on hybrid systems theory, optimization, and perception, will be created for dynamical models of legged locomotion in complex environments. The project will address the synthesis of robust-optimal controllers, based on nonlinear control and matrix inequalities, for agile and dexterous locomotion. It will also create safety-critical control algorithms, based on set invariance and quadratic programming, for real-world implementation of robust controllers that guarantee obstacle avoidance in unknown environments. To bridge the gap between theory and implementation, this research will transfer the theoretical innovations into practice through experiments with two advanced quadrupedal and bipedal robots in researchers' laboratories at Virginia Tech and Caltech sites.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.