NRI: Small: Collaborative Research: Rethinking Motion Generation for Robots Operating in Human Workspaces

NRI：小型：协作研究：重新思考在人类工作空间中操作的机器人的运动生成

• 批准号: 1317849
• 负责人: Lydia Kavraki
• 金额: $ 89.92万
• 依托单位: William Marsh Rice University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2013
• 资助国家: 美国
• 起止时间: 2013-09-01 至 2019-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1317849&HistoricalAwards=false
• 关键词: NRI; Small; Collaborative; Research; Rethinking

This proposal develops a new motion planning paradigm for enabling robots to work in the presence of humans or cooperatively with humans as co-workers. The paradigm is intended to be fast, reactive, and responsive to the requirements that arise from human-robot interaction. It involves the definition and subsequent implementation of constraints that encode properties of human-aware paths and can be translated to cost functions characterizing path quality. New motion planners are proposed. The operation of these planners is guided by constraints and their corresponding cost functions. The planners produce paths compatible with a given set of constraints. A mechanism to incorporate user feedback on the relative importance of constraints is provided and semi-autonomous operation of the robots is considered. Importantly, the planners are embedded in a novel hybrid-systems framework that allows a robot to automatically switch among planners, and hence behaviors, in order to take into account different constraints and user preferences that arise in the context of semi-autonomous operation. Besides the actual implementation of the planners on specific platforms, this project also disseminates all developed libraries and planners. Compatibility will the Robot Operating System (ROS) is provided for wide adoption, while tutorials at major conferences are planned. The training of graduate students and female undergraduate students are a central focus of this proposal.

该提案开发了一种新的运动规划范例，使机器人能够在人类面前工作或作为同事与人类合作。该范式旨在快速、反应灵敏，并能响应人机交互中产生的要求。它涉及约束的定义和后续实现，这些约束对人类感知路径的属性进行编码，并且可以转换为表征路径质量的成本函数。提出了新的运动规划器。这些规划器的操作受约束及其相应的成本函数的指导。规划器生成与给定的一组约束兼容的路径。提供了一种结合用户对约束相对重要性的反馈的机制，并考虑了机器人的半自主操作。重要的是，规划器嵌入了一个新颖的混合系统框架中，该框架允许机器人在规划器之间自动切换，从而自动切换行为，以便考虑半自主操作背景下出现的不同约束和用户偏好。除了在特定平台上实际实施规划器外，该项目还传播所有开发的库和规划器。提供了机器人操作系统（ROS）的兼容性以供广泛采用，同时计划在主要会议上提供教程。研究生和女本科生的培养是该提案的重点。