RI: Medium: Collaborative Research: Real-Time Continuum Manipulation

该项目的目标是研究在结构不确定甚至未知的低结构环境中使用新型连续体机器人的全新机器人操作模式的潜力。连续体机器人，如躯干/触手机械臂，在许多意义上是双重的。*其固有的灵活性和顺应性提供了更大的希望，使不精确和不确定条件下的物体在更大范围（数量级）的尺寸和重量，许多不同的形状，并具有广泛不同的物理特性（刚性，柔软，柔性等）灵巧的操作。另一方面，其固有的精度不足使得传统的机器人机械手规划和控制方法不适用于连续体机器人的操作。该项目开创了对单个连续体机器人自主操作具有许多不确定性的物体的基本问题的研究。它引入了一种新颖的整体方法，在实时传感和大环境不确定性下集成了实时自适应规划和鲁棒控制方案。接下来，它扩展了基本方法，以解决在公共环境中工作的多个连续体机器人，其中每个机器人不需要知道另一个机器人的运动。该研究将理论/算法开发与在配有传感器的真实躯干/触手机器人的实验测试台上进行的实际验证相结合。结果将通过出版物、自由软件和现实世界的演示积极传播，以影响研究、教育和应用程序。

The goal of this project is to investigate the potential of fundamentally new modes of robotic manipulation using novel continuum robots in less-structured environments with large uncertainties and even unknowns. A continuum robot, such as a trunk/tentacle robot arm, is in many senses ?dual? to a traditional robot manipulator (consisting of an articulated arm and a gripper or hand), featuring relatively low precision but high compliance: * Its inherent flexibility and compliance offer greater promise to enable deft manipulation under imprecise and uncertain conditions of objects over a wider range (orders of magnitude) of size and weight, of many different shapes, and with widely different physical characteristics (rigid, soft, flexible, etc.).. * On the other hand, its inherent lack of precision renders the traditional approaches to planning and control of robot manipulators unsuitable for manipulation with continuum robots. This project pioneers the study of the basic problem of autonomous manipulation of an object with much uncertainty by a single continuum robot. It introduces a novel and holistic approach that integrates real-time adaptive planning and robust control schemes under real-time sensing and large environmental uncertainty. It next extends the basic approach to address multiple continuum robots working in a common environment, where each robot needs not know the motion of another robot. The research combines theoretical/algorithmic development with real-world validation on an experimental test bed with real trunk/tentacle robots equipped with sensors. The results will be actively disseminated through publications, free software, and real-world demos to impact research, education, and applications.