S&AS: INT: Socially-Aware Autonomy for Long-Term Deployment of Always-On Heterogeneous Robot Teams

S

• 批准号: 1724157
• 负责人: Scott Niekum
• 金额: $ 110万
• 依托单位: University of Texas at Austin
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-09-01 至 2022-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1724157&HistoricalAwards=false
• 关键词: S& INT; Socially; Aware; Autonomy

This project studies how best to enable a team of mobile robots to fully integrate into the social fabric of a community of people. As such, the most important impact of this project is the ongoing development and dissemination of core algorithms that, for the first time, enable robots to be fully integrated in a socially-aware way into a community of people over a long period of time. In addition to many traditional impacts on education, including a curriculum for a project-based course, serving as a platform for graduate and undergraduate research, this research is being used as the basis for extensive public outreach via demonstrations, camps, and short courses, with a particular focus on underrepresented groups such as women in Computer Science and Engineering and Hispanic students.This project focusses on three main technical challenges that are crucial to enabling robots to integrate into the social fabric of a community. First, the project is working towards enabling robots to be cognizant about dynamic social environments. Second, the project is working towards enabling robots to reason about heterogeneous human-robot capabilities while carrying out specific tasks. Third, the project is working towards enabling safe, reflective lifelong learning in dynamic environments. In addition to leading to focussed algorithms designed to address these specific challenges, the project contributions are also being instantiated within a single, coherent multi-robot system capable of long-term autonomy. A key facilitator of the project's validation plan is a flexible, adaptable, and reusable software and hardware infrastructure that is being developed and expanded at UT Austin under an NSF Computing Research Infrastructure program grant. The main experimental objective of the project is to demonstrate that these robots can not only operate autonomously for long periods of time, but also be fully integrated into and accepted by the community of people who use the building in which they are situated.

该项目研究如何最好地使移动的机器人团队完全融入社区的社会结构。 因此，该项目最重要的影响是核心算法的持续开发和传播，这些算法首次使机器人能够以具有社会意识的方式在很长一段时间内完全融入人类社区。除了对教育的许多传统影响外，包括作为研究生和本科生研究平台的基于项目的课程的课程，这项研究还被用作通过示范，营地和短期课程进行广泛公众宣传的基础，特别关注代表性不足的群体，如计算机科学与工程领域的女性和西班牙裔学生。这些挑战对于使机器人融入社区的社会结构至关重要。首先，该项目致力于使机器人能够认识到动态的社会环境。 其次，该项目正在努力使机器人能够在执行特定任务时推理异质的人机能力。 第三，该项目致力于在动态环境中实现安全、反思性的终身学习。 除了导致旨在解决这些特定挑战的集中算法外，项目贡献还在能够长期自主的单个连贯多机器人系统中实例化。 该项目验证计划的一个关键推动者是一个灵活、适应性强、可重复使用的软件和硬件基础设施，该基础设施正在德克萨斯大学奥斯汀分校根据NSF计算研究基础设施计划拨款进行开发和扩展。 该项目的主要实验目标是证明这些机器人不仅可以长时间自主运行，而且还可以完全融入并被使用它们所在建筑物的社区所接受。

项目成果

PixL2R: Guiding Reinforcement Learning Using Natural Language by Mapping Pixels to Rewards

• 发表时间: 2020-07
• 期刊: ArXiv
• 作者: Prasoon Goyal;S. Niekum;R. Mooney

SCAPE: Learning Stiffness Control from Augmented Position Control Experiences

SCAPE：从增强的位置控制经验中学习刚度控制

• 发表时间: 2021
• 期刊: Conference on Robot Learning
• 作者: Kim, M;Niekum, S;Deshpande, A
• 通讯作者: Deshpande, A

Better-than-Demonstrator Imitation Learning via Automatically-Ranked Demonstrations

• 发表时间: 2019-07
• 作者: Daniel S. Brown;Wonjoon Goo;S. Niekum

A Review of Robot Learning for Manipulation: Challenges, Representations, and Algorithms

• 发表时间: 2019-07
• 期刊: J. Mach. Learn. Res.
• 作者: Oliver Kroemer;S. Niekum;G. Konidaris

Human Gaze Assisted Artificial Intelligence: A Review

• DOI: 10.24963/ijcai.2020/689
• 发表时间: 2020-07
• 期刊: IJCAI : proceedings of the conference
• 作者: Ruohan Zhang;Akanksha Saran;Bo Liu-;Yifeng Zhu;Sihang Guo;S. Niekum;D. Ballard;M. Hayhoe