RI: Small: Functional Object-Oriented Network for Manipulation Learning

RI：小型：用于操作学习的功能性面向对象网络

• 批准号: 1421418
• 负责人: Yu Sun
• 金额: $ 39.85万
• 依托单位: University of South Florida
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-08-15 至 2019-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1421418&HistoricalAwards=false
• 关键词: RI; Small; Functional; Object; Oriented

This project constructs a functional object-oriented network (FOON) to model the connectivity of the functional-related objects and their functional motions in manipulation tasks. The graphical model FOON are developed and learned by observing state change of objects and human manipulations with the objects. A well-trained FOON enables robots to decipher a task goal, seek the correct objects at the desired states, and generate an optimal motion, which eventually allows robots to perform daily living tasks and provide needed help to people with disabilities and seniors. The project builds a kitchen environment with representative kitchen objects and a virtual environment to collect training data and learn a representative FOON from the observations of self-generated manipulation tasks performed in both environments. The data collected in the proposed work enables researchers to fully explore daily living tasks from the object state and object motion point of view and provide excellent training and testing data sets for other related research. The FOON provides a powerful tool for discovering new insights in neuroscience, cognitive science, and psychology. It is also useful in training to break down skills for teaching and training evaluation. This project draws additional minority students through the exceptional diversity programs that exist at USF. The project supports a summer camp to attract young students to STEM careers and the research results are widely disseminated through publications and inclusion in courses.

该项目构建了一个面向功能对象的网络(FOON)来模拟操作任务中与功能相关的对象及其功能运动的连通性。图形模型FOON是通过观察对象的状态变化和人类对对象的操纵来开发和学习的。训练有素的FOON使机器人能够破译任务目标，在期望的状态下寻找正确的物体，并产生最佳运动，最终使机器人能够执行日常生活任务，并为残疾人和老年人提供必要的帮助。该项目建立了一个具有代表性的厨房对象的厨房环境和一个虚拟环境，以收集训练数据，并从在这两个环境中执行的自我生成的操纵任务的观察中学习具有代表性的FOON。所收集的数据使研究人员能够从物体状态和物体运动的角度全面探索日常生活任务，并为其他相关研究提供良好的训练和测试数据集。FOON为发现神经科学、认知科学和心理学的新见解提供了一个强大的工具。在培训中，分解技能也是有用的，用于教学和培训评估。该项目通过USF现有的特殊多样性项目吸引了更多的少数族裔学生。该项目支持举办夏令营，以吸引年轻学生投身STEM职业，并通过出版物和课程内容广泛传播研究成果。