
NSF Org: |
IIS Division of Information & Intelligent Systems |
Recipient: |
|
Initial Amendment Date: | July 25, 2017 |
Latest Amendment Date: | March 19, 2019 |
Award Number: | 1718478 |
Award Instrument: | Standard Grant |
Program Manager: |
Juan Wachs
IIS Division of Information & Intelligent Systems CSE Directorate for Computer and Information Science and Engineering |
Start Date: | August 15, 2017 |
End Date: | July 31, 2022 (Estimated) |
Total Intended Award Amount: | $500,000.00 |
Total Awarded Amount to Date: | $508,000.00 |
Funds Obligated to Date: |
FY 2019 = $8,000.00 |
History of Investigator: |
|
Recipient Sponsored Research Office: |
6100 MAIN ST Houston TX US 77005-1827 (713)348-4820 |
Sponsor Congressional District: |
|
Primary Place of Performance: |
6100 Main St Houston TX US 77005-1827 |
Primary Place of
Performance Congressional District: |
|
Unique Entity Identifier (UEI): |
|
Parent UEI: |
|
NSF Program(s): | Robust Intelligence |
Primary Program Source: |
01001920DB NSF RESEARCH & RELATED ACTIVIT |
Program Reference Code(s): |
|
Program Element Code(s): |
|
Award Agency Code: | 4900 |
Fund Agency Code: | 4900 |
Assistance Listing Number(s): | 47.070 |
ABSTRACT
Motion planning is the problem of determining how to get the robot from one point to another. Ideally, robots should have past experiences, of their own and others, inform future actions to operate more robustly and improve their performance over time. Motion planning, as it is largely practiced today, focuses on solving one problem at a time and makes limited use of past history. The goal of this project is to transform the way robots plan their motions by learning to exploit similarities between different experiences and by creating strategies that can adapt to wide range of scenarios. The work will create a bridge between the motion planning community and the information retrieval community, potentially transforming both fields. Training opportunities for diverse students will be offered. All developed software is disseminated under an open source license and infrastructure will enable other researchers to use the experience databases and contribute to them.
This project provides a two-pronged approach to transform motion planning using an experience database. First, hashing will be used on an environment to fetch roadmaps for similar environments from a database. A roadmap is a graph representing feasible motions for a robot. These fetched roadmaps will be then lazily composed and refined to allow the robot to plan efficiently in the current environment. The use of prior experience will be done in tandem with planning from scratch; the latter, if successful, can provide a path and add to the experience database. The second prong in the planned approach will be to maintain various performance characteristics of a library of motion planning algorithms. These characteristics will be then used to optimize algorithm performance and construct a portfolio of algorithms that is competitive across various problems. The overall framework will be implemented in the cloud.
PUBLICATIONS PRODUCED AS A RESULT OF THIS RESEARCH
Note:
When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external
site maintained by the publisher. Some full text articles may not yet be available without a
charge during the embargo (administrative interval).
Some links on this page may take you to non-federal websites. Their policies may differ from
this site.
PROJECT OUTCOMES REPORT
Disclaimer
This Project Outcomes Report for the General Public is displayed verbatim as submitted by the Principal Investigator (PI) for this award. Any opinions, findings, and conclusions or recommendations expressed in this Report are those of the PI and do not necessarily reflect the views of the National Science Foundation; NSF has not approved or endorsed its content.
As robots enter our daily lives, they need to become multi-purpose and multi-functional and improve their motion planning capabilities in diverse scenarios, while, at the same time, allowing their specialization and robust operation in a certain type of environment. A service robot for example, may specialize in operating efficiently in an apartment computing and executing reliable and optimized motion, while another robot of the same type may become an efficient sorter of small mechanical parts that arrive in boxes. Ideally, robots should have past experiences inform future actions to operate more robustly and improve their performance over time. Motion planning, as it is often understood and practiced, focuses on solving one problem at a time and makes limited use of history. This proposal transforms the way robots plan their motions by learning to exploit similarities between motion planning problems and by creating adaptive algorithmic strategies that are robust across a wide range of scenarios. The intellectual merit lies in new framework for learning sampling distributions that guide the most successful category of today’s motion planners, namely sampling-based motion planners. By combining non-parametric learning methods (based on information retrieval) and local geometric decompositions, the resulting framework successfully uses earlier experiences stored in a database resulting in an order of magnitude improvement over competing learning methods. The developed methodology also shows how to learn retrieval functions, specific for motion planning problems, utilizing contrastive learning and a self-supervised training scheme. The approach has been successful with modern manipulators operating in realistic environments obtained through sensing. While investigating the challenges of integrating planning algorithms with modern machine learning methods, this proposal has also developed a path deformation method which can generalize to several scenarios from a single demonstration, a method that uses past experiences for constrained motion planning, differential surrogates for (nondifferential) planners to optimize the respective learned sampling distributions, and a methodology for tuning hyper-parameters of planners. It has also contributed tools for planner evaluation. Evaluating new planners remains a challenge, as researchers often create their own ad-hoc benchmarking problems, a process which is time consuming and bias prone. The advent of learning-based planners has exacerbated this problem as such planners rely on datasets both for learning and evaluation. Towards creating community accepted datasets and benchmarks, a benchmarking dataset generator was proposed along with several challenging motion planning problems. The broader impact of this proposal has been realized by tight collaborations between a machine learning and a robotics group, by training both undergraduate and graduate students, by distributing open-source software, and by introducing the developed concepts in an undergraduate class at Rice University.
Last Modified: 11/13/2022
Modified by: Lydia Kavraki
Please report errors in award information by writing to: awardsearch@nsf.gov.