NRI: Collaborative Research: Human-Supervised Manipulation of Deformable Objects

提出的工作目标是开发算法，使人类监督的机器人在重大不确定性下操纵可变形物体。可变形物体的操作在外科手术中是必不可少的，这涉及到在很大的不确定性下对精致和高度可变形组织的复杂操作，以及在制造业中，许多装配任务涉及柔性部件和材料（例如电缆，纺织品和复合材料）。机器人硬件的最新进展（例如达芬奇和巴克斯特机器人）使得机器人对可变形物体的操作在物理上成为可能，但是当存在显著的模型不确定性时，机器人仍然缺乏在该领域执行实际任务所需的算法。该项目将通过其在外科和制造机器人中的应用产生广泛的社会影响。鲁棒操纵可变形结构的能力是实现智能机器人手术助手的重要先导技术。机器人技术及其医疗应用有可能激励孩子们在STEM领域从事职业，并满足美国不断增长的医疗保健和制造机器人行业的需求。研究活动与教育的整合将强调积极地让本科生参与研究活动，并在本科和研究生课程中引入新的讲座材料和项目。此外，将特别强调从代表性不足的群体中招收合格的学生。提出的研究目的是开发算法，使人类监督的机器人能够在很大的不确定性下操纵可变形的物体。具体而言，研究将侧重于开发用于建模可变形物体动力学和相关不确定性的自适应复杂性模型，用于集成探索和任务执行的规划算法，用于不确定性下可变形物体鲁棒操作的控制算法，以及用于有效监督机器人操作可变形物体的算法。该项目的智力价值来自于它对机器人建模、规划和控制算法的基本贡献，这些算法用于操纵可变形物体。该研究将对机器人操纵、不确定条件下的运动规划、柔性操纵控制等领域产生重要影响。

The goal of the proposed work is to develop algorithms that enable human-supervised robotic manipulation of deformable objects under significant uncertainty. Manipulation of deformable objects is essential in surgery, which involves complex manipulations of delicate and highly deformable tissue under substantial uncertainty, and in manufacturing, where many assembly tasks involve flexible parts and materials (e.g. cables, textiles, and composites). Recent advances in robot hardware (e.g. the da Vinci and Baxter robots) have made robotic manipulation of deformable objects physically possible but robots still lack the algorithms necessary to perform practical tasks in this domain when there is significant model uncertainty. This project will have broad societal impact through its applications in surgical and manufacturing robotics. The ability to robustly manipulate deformable structures is an important precursor technology towards realizing intelligent robotic surgical assistants. Robotics and its medical applications have the potential to inspire children to pursue careers in STEM fields and meet the needs of America's growing health-care and manufacturing robotics industry. Integration of the research activities with education will emphasize actively involving undergraduates in research activities and introducing new lecture material and projects into undergraduate and graduate courses. Also, special emphasis will be given to recruit qualified students from under-represented groups. The purpose of the proposed research is to develop algorithms that enable human-supervised robotic manipulation of deformable objects under substantial uncertainty. Specifically, the research will focus on developing adaptive complexity models for modeling deformable object dynamics and associated uncertainty, planning algorithms for integrated exploration and task execution, control algorithms for robust manipulation of deformable objects under uncertainty, and algorithms for effective human supervision of robotic manipulation of deformable objects. The intellectual merit of the project comes from its fundamental contributions to robotic modeling, planning, and control algorithms for manipulation of deformable objects. The research will have impact in the fields of robotic manipulation, motion planning under uncertainty, and compliant manipulation control.