Award Abstract # 1739315
CPS: Medium: Enabling Real-time Dynamic Control and Adaptation of Networked Robots in Resource-constrained and Uncertain Environments

NSF Org: CNS
Division Of Computer and Network Systems
Recipient: RUTGERS, THE STATE UNIVERSITY
Initial Amendment Date: August 14, 2017
Latest Amendment Date: April 25, 2019
Award Number: 1739315
Award Instrument: Standard Grant
Program Manager: David Corman
dcorman@nsf.gov
 (703)292-8754
CNS
 Division Of Computer and Network Systems
CSE
 Direct For Computer & Info Scie & Enginr
Start Date: September 1, 2017
End Date: August 31, 2021 (Estimated)
Total Intended Award Amount: $999,904.00
Total Awarded Amount to Date: $1,015,904.00
Funds Obligated to Date: FY 2017 = $999,904.00
FY 2019 = $16,000.00
History of Investigator:
  • Dario Pompili (Principal Investigator)
    pompili@rutgers.edu
  • Jingang Yi (Co-Principal Investigator)
  • Francisco Diez-Garias (Co-Principal Investigator)
Recipient Sponsored Research Office: Rutgers University New Brunswick
3 RUTGERS PLZ
NEW BRUNSWICK
NJ  US  08901-8559
(848)932-0150
Sponsor Congressional District: 12
Primary Place of Performance: Rutgers University New Brunswick
94 Breet Road
Piscataway
NJ  US  08854-3925
Primary Place of Performance
Congressional District:
06
Unique Entity Identifier (UEI): M1LVPE5GLSD9
Parent UEI:
NSF Program(s): S&AS - Smart & Autonomous Syst,
Special Projects - CNS,
CSR-Computer Systems Research,
CPS-Cyber-Physical Systems
Primary Program Source: 01001718DB NSF RESEARCH & RELATED ACTIVIT
01001920DB NSF RESEARCH & RELATED ACTIVIT
Program Reference Code(s): 046Z, 7354, 7918, 7924, 9251
Program Element Code(s): 039Y, 1714, 7354, 7918
Award Agency Code: 4900
Fund Agency Code: 4900
Assistance Listing Number(s): 47.070

ABSTRACT

Near-real-time water-quality monitoring in rivers, lakes, and water reservoirs of different physical variables is critical to prevent contaminated water from reaching the civilian population and to deploy timely solutions, or at least to issue early warnings so as to prevent damage to human and aquatic life. In order to make optimal decisions and "close the loop" promptly, it is necessary to collect, aggregate, and process water data in real time. Therefore, the goal of this project is to design a Cyber Physical System (CPS) where drones such as the Rutgers multi-medium Naviator, a Hybrid Unmanned Air/Underwater Vehicle (HUA/UV), and autonomous underwater robots (e.g., modified BlueROVs) can (i) first identify Regions of Interest (RoIs) and take measurements and well as, if needed, collect biosamples from them; (ii) and then, through collaborative information fusion and integration, perform in-situ transformation of these measurements/raw data into valuable information and, finally, into knowledge. To achieve the above goal, this project will need to solve the problem of uncertainties that arise in in-situ processing of data from sensors in any CPS. This project will provide greater autonomy and cooperation in CPSs and, at the same time, will improve scalability, reliability, and timeliness in comparison to traditional sensing systems. The challenges to achieve dynamic collaboration between local and cloud resources will be handled in Task 1, in which novel adaptive-sampling solutions that minimize the sampling cost of a RoI (in terms of time or energy expenditure) will also be developed. In Task 2, novel solutions will be designed to handle model uncertainties in the local resources due to the unpredictable behavior of computational models to input data and resources' availability. In Task 3, the project aims at developing a biosampler, i.e., "lab-on-robot", that uses in-situ measurements and communicates with the cloud resources to give results in real time on the water quality; also, new solutions to optimize the Naviator's current hybrid air/water multirotor platform/propulsion system will be designed in order for it to be able to carry and perform testing with the biosampler while also increasing its endurance. Finally, in Task 4, integrated field testing on the Raritan River, NJ, will be performed so as to validate the algorithms as well as to analyze their scalability (from an economical and feasibility perspective) and confidence/accuracy performance. Specifically, the Naviators will identify the RoIs via multimodal operations, i.e., in water and air; and then the BlueROVs (which, during the course of the project, will be made autonomous and will be modified to carry on-board water-quality sensors) will perform underwater adaptive sampling in each of those RoIs using the algorithms designed in Task 1.

In terms of broader impacts, the collaboration between cloud and local resources can benefit any CPS in the following ways: (i) outsourcing computation to the cloud will allow resource-constrained vehicles (in terms of computational capability) to meet mission deadlines, and (ii) using clouds comes at a price, hence, in order to accomplish the mission goals within budget constraints, the computational tasks composing a workflow should be migrated from the local network to the cloud only when the former does not have enough computational resources to execute successfully the tasks (outbursting). In terms of outreach, this project will develop a pipeline of diverse and computer literate engineers who will be able to solve self-management CPS problems. The PIs will 1) create a course on real-time in-situ distributed computing (for graduate computer engineering and undergraduate non-engineering majors); 2) develop teaching modules for incorporation into key high-school activities; 3) leverage existing minority student outreach programs and networks at Rutgers; 4) incorporate exchange programs and team-teaching approaches; and 5) utilize distributed education technologies with application to robotics and networking. Our electrical/computer and mechanical engineering team has the theoretical and system-level skills, cross-disciplinary expertise, as well as a verifiable history of fruitful collaboration to exploit fully this project's research and educational potential.

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.

Rahmati, Mehdi and Nadeem, Mohammad and Sadhu, Vidyasagar and Pompili, Dario "UW-MARL: Multi-Agent Reinforcement Learning for Underwater Adaptive Sampling using Autonomous Vehicles" ACM International Conference on Underwater Networks and Systems (WUWNet) , 2019 10.1145/3366486.3366533 Citation Details
Chen, Wenjie and Rahmati, Mehdi and Sadhu, Vidyasagar and Pompili, Dario "Real-time Image Enhancement for Vision-based Autonomous Underwater Vehicle Navigation in Murky Waters" ACM International Conference on Underwater Networks and Systems (WUWNet) , 2019 10.1145/3366486.3366523 Citation Details
Rahmati, Mehdi and Arjula, Archana and Pompili, Dario "Compressed Underwater Acoustic Communications for Dynamic Interaction with Underwater Vehicles" ACM International Conference on Underwater Networks and Systems (WUWNet) , 2019 10.1145/3366486.3366488 Citation Details

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.

This project designed a novel Cyber Physical System (CPS) architecture for near-real-time water monitoring using heterogeneous and autonomous sensing entities/vehicles, where drones such as the Rutgers multi-medium Naviator, a Hybrid Unmanned Air/Underwater Vehicle (HUA/UV), and autonomous underwater robots (e.g., modified BlueROVs) can (i) first identify Regions of Interest (RoIs) and take measurements as well as, if needed, collect biosamples from them; (ii) and then, through collaborative information fusion and integration, perform in-situ transformation of these measurements/raw data into valuable information and, finally, into knowledge. The project provided greater autonomy and cooperation in CPSs and improved scalability, reliability, and timeliness in comparison to traditional sensing systems.

In terms of intellectual merit, this project achieved three major goals: (i) performing dynamic and near-real-time collaboration between local and cloud resources; (ii) designing novel solutions to handle model uncertainties in the local resources due to the unpredictable behavior of computational models to input data and resources? availability; (iii) performing integrated field testing on the Raritan River, NJ so as to validate the designed algorithms as well as to analyze their scalability (from an economical and feasibility perspective) and to assess their confidence and accuracy performance.

In terms of education outreach, besides supporting several graduate students (MS and PhD) throughout the four years of the project, this collaborative effort exposed undergraduate (UG) students to research by having them investigate the proposed CPS sensing architecture for near-real-time water monitoring using heterogeneous and autonomous sensing entities/vehicles. Many UG students contributed to building, debugging, and testing the proposed sensing system, which consists of three main parts: (i) CPU-based computational units, (ii) sensor-based data collection units, and (iii) semi-autonomous robotic platforms.

Specific achievements/outcomes accomplished during this project are summarized below.

  • Dynamic interaction with underwater vehicles using acoustic communications. We conducted simulations as well as extensive experiments using an autonomous vehicle with WHOI micro-modems in the Raritan River, Somerset, Carnegie Lake in Princeton, and in the Marine Park in Red Bank, all in New Jersey. We modified an underwater ROV through the installation of a WHOI acoustic micro-modem to enable semi-autonomous missions for the vehicle through acoustic communications with a control center. In the proposed communications protocol, both vehicle and the center/surface hub station keep silent when the measured data matches the prediction so as to use efficiently the limited bandwidth underwater and for added robustness.
  • Source feature compression for object classification in underwater robotics. We tested the proposals with the underwater object data that was captured from the Raritan River in New Jersey. It is verified that the proposals effectively reduce the training time for the underwater object classification task and improve the validation accuracy convergence compared with competing methods. The object classification is an essential part of an underwater robot that can sense the underwater environment and perform autonomous navigation. Therefore, the proposal is well-suited for efficient computer vision-based tasks in underwater robotic applications.
  • Multi-agent hierarchical reinforcement learning for underwater adaptive sampling using autonomous vehicles. We proposed a MAHRL ?hierarchical? framework to make efficient sequence of decisions for underwater adaptive sampling using autonomous vehicles. A map reconstruction algorithm and a communication protocol were proposed to achieve the distributed workload. The solution was evaluated via computer simulations (to find optimal values for different design parameters) and was shown to achieve the desired performance. In the simulation engine, we defined the environment and a variable number of agents. For each agent, we had a prescribed depth.
  • Adding robustness to our autonomous underwater robots. Currently, we have operationalized the BlueROV using the wireless as well as wired connections. We work towards implementing a closed-loop control strategy where the ROV is able to correct automatically its trajectory dynamically in the face of external disturbances such as currents, winds, and waves. For that, we have looked into the mathematical model for the ROV and hydrodynamics, the model for the controller, filtering methods for sensor data such as Kalman Filters, and underwater simulation setup.
  • Cross-layering strategies in underwater acoustic networks for scalable video transmissions. We proposed a novel adaptive cross-layer video transmission protocol for underwater acoustic networks, namely CLPSVT, to improve the system robustness as well as to meet the requirements of video quality. Both the transmitter and receivers utilize cross-layer interactions between the physical and the application communication layers to select the scalable video encoding/decoding method and the modulation and coding scheme adaptively.
  • Circular time shift modulation for underwater acoustic communications. We proposed a novel physical-layer modulation, called CTSM, to improve the system robustness when in underwater acoustic channels with multipath and additive noise. The key feature of the proposed modulation is the utilization of the high auto-correlation property of zero-correlation-zone signals. The CTSM was further validated under emulations with real underwater acoustic channels extracted using the NATO CMRE LOON testbed in La Spezia, Italy. 

 


Last Modified: 01/18/2022
Modified by: Dario Pompili

Please report errors in award information by writing to: awardsearch@nsf.gov.

Print this page

Back to Top of page