RI: Small: Sampling Based Feedback Motion Planners

RI：小型：基于采样的反馈运动规划器

• 批准号: 1217991
• 负责人: Suman Chakravorty
• 金额: $ 36.92万
• 依托单位: Texas A&M Engineering Experiment Station
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2012
• 资助国家: 美国
• 起止时间: 2012-09-01 至 2016-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1217991&HistoricalAwards=false
• 关键词: RI; Small; Sampling; Based; Feedback

This project, developing robust and computationally intelligent randomized sampling based feedback motion planning techniques for constrained systems operating under "process" and "sensing" uncertainty, addresses an important research area for real-world systems. Such motion planning techniques are valuable because the high computational burden of these problems makes the solution of such problems intractable for anything but the simplest low dimensional systems. In particular, one of the most fundamental requirement of robots is that they operate in a safe, efficient and autonomous fashion in the presence of such uncertainty. Thus, a principled set of computationally intelligent techniques, with guaranteed performance, is required. This proposed work will generalize the Probabilistic RoadMap technique (PRM) of robotic path planning shall such that the roadmap construction incorporates both process and sensing uncertainty. This will result in a computationally tractable solution technique for a large class of constrained Markov Decision Problems (MDP) and Partially Observed MDPs (POMDP), known as constrained stochastic shortest path problems, along with guaranteed performance of the planners in terms of a probability of success. Broader Impacts: The ability to solve high dimensional constrained MDP and POMDP problems in a computationally tractable fashion will have significant impact on multiple different robotic applications including robotic operations in hazardous environments such as disaster areas and battlefields, surgical robotics, prosthetics and unmanned planetary exploration. The impact of such computationally intelligent solution techniques cannot be overstated due to the ubiquitous nature of MDPs and POMDPs, which are fundamental decision making problems inherent in myriad different fields ranging from Engineering through Economics to Biology. The assimilation of K-12/undergraduate/graduate students, with a focus on underrepresented minorities, with high school teachers in projects related to the research, through experiments and demonstrations related to the research at the annual "TAMU Physics and Engineering fair", and the annual department "summer camp" for high school students and their parents, will disseminate the results of the project to a broad audience.

该项目，开发强大的和计算智能的随机采样为基础的反馈运动规划技术的约束系统下运行的“过程”和“传感”的不确定性，解决了一个重要的研究领域，为现实世界的系统。这样的运动规划技术是有价值的，因为这些问题的高计算负担使得这些问题的解决方案对于除了最简单的低维系统之外的任何东西都是棘手的。特别是，对机器人最基本的要求之一是，它们在存在这种不确定性的情况下以安全、高效和自主的方式运行。因此，需要一套有原则的计算智能技术，保证性能。这项拟议的工作将概括机器人路径规划的概率路线图技术（PRM），应使路线图建设结合了过程和传感的不确定性。这将导致在一个计算上易于处理的解决方案技术的一大类的约束马尔可夫决策问题（MDP）和部分观察MDP（POMDP），被称为约束随机最短路径问题，沿着与保证性能的规划成功的概率。更广泛的影响：以计算易处理的方式解决高维约束MDP和POMDP问题的能力将对多种不同的机器人应用产生重大影响，包括在危险环境中的机器人操作，如灾区和战场，外科手术机器人，假肢和无人行星探索。这种计算智能解决方案技术的影响不能被夸大，由于MDP和POMDP的无处不在的性质，这是从工程到经济学到生物学的无数不同领域中固有的基本决策问题。通过在每年一度的“TAMU物理和工程博览会”上进行与研究有关的实验和示范，以及每年一度的高中生及其家长系“夏令营”，将K-12/本科生/研究生（重点是代表性不足的少数民族）与高中教师同化，将向广大受众传播项目的成果。