RiverNet - Distributed Sensor Nets for Environmental Monitoring
Written by Arthur Sanderson (RPI) & D. Richard Blidberg (AUSI)
Monday, 15 June 2009
During 2008, the RiverNet Project has focused on the development of algorithms for multivehicle navigation and cooperative behaviors. Solar-power AUV (SAUV) technology offers the capability of long-term deployment of undersea vehicles for observation and monitoring of physical, chemical, and biological features and events with application to a variety of security, defense, and environmental missions. In this approach, a SAUV vehicle is required to surface for periods to recharge, then can resubmerge. While a single vehicle can carry out such a mission, the use of multiple vehicles (typically three or more) extends the capabilities in two ways [23]. First, at least one vehicle may be submerged at all times, providing continuous monitoring. Second, the vehicles may be dynamically reconfigured to improve navigation capability of the group.{mosgoogle}
In this phase of the research, we have developed both Kalman and particle filtering algorithms as a basis for improved navigation of multiple SAUV vehicles. Simulation analysis of these algorithms demonstrates that improved navigation capability may be achieved by multivehicle cooperation. In addition, the particle filter offers some advantages over the Kalman filter for these applications where non-linear relationships and non-Gaussian probability models may hold. However, the particle filter requires additional computational power and time. Therefore, a hybrid algorithm integrating both Kalman and particle filter components is being studied.
The development of distributed sensor networks integrating mobile autonomousunderwater robotic vehicles has continued and new applications to problems ofenvironmental monitoring with special emphasis on the Hudson River and Estuary have been demonstrated. Algorithms and technology have been developed within theRiverNet project with the overall goal of analysis and design of distributed sensornetworks for observing data from complex and geographically extended regions. Thesethree principal activities are:
1. Distributed spatial and temporal sampling of distributed variable fields incorporatingfixed and mobile sensor nodes. Significant progress has been achieved on the design ofparametric and non-parametric methods and these are being used in both simulation andexperimental testing of distributed sensor systems.{mosgoogle}
2. Research conducted in conjunction with The Beacon Institute will lead to extensionsto the deployment of a Hudson River sensing network that provides physical andchemical sensing variables on the Hudson River and retrieves data by wireless connections. This sensor network will include conductivity, temperature, depth,turbidity, chlorophyll, and dissolved oxygen. This effort is being further augmented by a ending partnership with the IBM Corporation.
3. Progress in the development and deployment of the Solar autonomous underwater ehicle (SAUV) in a series of test phases is designed to validate vehicle performance and evelop reliable mission planning and navigation capability. As of 2008, five SAUV ehicles have been manufactured, and further testing is planned over the next 12 months o demonstrate the capabilities of the vehicles for sensing, communications, and ooperative navigation. These results and test activities are described in more detail elow. Cooperating organizations involved in the work described include Rensselaer olytechnic Institute (RPI), AUSI, Technology Systems, Inc. (TSI), Falmouth Scientific, nc. (FSI), the Naval Undersea Warfare Center (NUWC)-Newport, The Beacon Institute and the University of New Hampshire (UNH).
1. SAUV Platform The SAUV, shown in Figure 1, is a solar-powered AUV designed for long endurance missions that require monitoring, surveillance or station keeping, with real time bidirectional communications to shore [1,2,3]. The SAUV is designed to operate as a normal AUV when submerged but will also reside on the surface while recharging batteries.
While on the surface, the SAUV is designed to communicate remotely via Iridium satellite or RF communications link to upload collected data and to allow remote reprogramming of mission profiles. A typical mission scenario would include submerged operation at night for about 12 hours and surface recharging and communication for about 12 hours during the daytime. This mission could last weeks to months with daily updates provided to the user from anywhere in the world. The vehicle can be preprogrammed to submerge to depths of 500 meters, to transit to designated waypoints, or to operate on the surface during conditions suitable for battery charging via solar energy.
Figure 1. Solar-powered AUV (SAUV) platform.
Figure 1. Solar-powered AUV (SAUV) platform.
The SAUV system employs a comprehensive and flexible set of capabilities for communications and remote data transfers. Real-time vehicle performance and state data are logged internally and are available through ftp transfer while the SAUV is on the surface. During a mission, payload, sensor and state data are available on a periodic or polled basis through the acoustic channel. The data typically is picked up by a gateway buoy (or gateway SAUV) and relayed via RF to the remote land site where it is displayed on a computer using the Mission Planner software application.
Mission Sensors The basic SAUV mission sensor suite includes sensors for altitude, pressure (depth), conductivity (salinity) and temperature. One of the key design goals in the development of the SAUV was as a platform to host various sensor systems. Recent sensor surveys have depended upon the collection of additional science data. For example, the MB06 science mission undertaken in Monterey Bay, CA in 2006 relied upon the installation of an OxyGuard 505 dissolved oxygen sensor and a WET Labs ECO Puck combined fluorometer/turbidity sensor, in addition to the standard FSI NXIC CTD sensor on each vehicle.
Hardware Development {mosgoogle}Several aspects of the SAUV system hardware have evolved during the period of performance of this NSF Program. The performance and stability of the vehicle have been heavily dependent on the placement of trim weights. Small changes in configuration require tedious re-trimming. Steps to address this issue have been undertaken which include increasing the vectored thruster shroud diameter and exploring the use of vertical fins, and are based upon a careful study of the hydrodynamics of the vehicle [4].
Another area where vehicle performance has been improved is in the Li-Ion battery subsystem. In 2007, transition from the Mathews battery system to another system built by SouthWest Electronic Energy Corp was completed. This new system has proven to be more reliable and much easier to handle, resulting in less operational down time.
Software Development The software capabilities of the SAUV platform have also evolved over the past several years, with changes primarily in two areas. In 2006, the capability to perform “system in the loop” testing of the SAUV high-level software was developed. Along with this addition came the capability to more easily plug in new mission sensors, enhancing the ability of the SAUV to be used in different kinds of sensor surveys.
The second area which has evolved is in the high-level control of the SAUV. Focus areas include design and implementation of a common control language (CCL) for AUVs and development of a distributed control environment (DICE) as well as high-level vehicle behaviors.
2. Tools and Technologies Supporting Multi-vehicle Cooperation A set of enabling tools and technologies which allow us to test and evaluate multiple cooperating AUVs has been developed. These technologies are critical to enabling a fleet of AUVs to perform cooperative tasks such as adaptive sampling and surveying. The following is a summary of those technologies which were partially funded by this NSF support.
SAUV “System in the Loop” Simulation During 2006, researchers augmented the Cooperative AUV Development Concept (CADCON) environment to provide “system in the loop” capability for testing and evaluating SAUV system components and multiple cooperative vehicle mission profiles before going in the water [5]. This simulation facility allows for complete testing of SAUV onboard high-level software, including underwater networking protocol logic. The facility also has a training functionality in that top level mission planning and vehicle monitoring applications used by SAUV operators can also be tested as if they were in a field setting. Hardware components, such as radio frequency (RF) and acoustic modems can also be tested within the systems context. In this harness, the SAUV PC-104 system, running a Linux OS and the high-level software, can be tested as a networked bench-level component. In addition, significant portions of the standalone SAUV can be put into simulation mode, thereby allowing the testing of other on-board vehicle electronics and subsystems.
Common Control Language (CCL) Multiple vehicle cooperation relies upon the ability of vehicles to communicate and understand each other. AUSI and NUWC are developing a CCL to provide (1) a common messaging interface to different AUVs, (2) an operator to vehicle group mission specification interface, (3) a sufficiently rich vocabulary and grammar to permit development of high level behaviors from lower level behaviors, and (4) support for optimization strategies for multiple AUV cooperation [6]. The message specification in particular draws heavily upon past work in “generic behaviors” [7] and other AUV command languages, as well as work done in intelligent agent communications. It is explicitly designed to support a wide range of vehicle types in its command and informational structure. In addition, this protocol allows for arbitrary execution of behaviors (parallel, sequential, adversary, general choice, cost choice) and, when combined with $-calculus, allows vehicles to accept a goal, jointly plan how to achieve that goal and carry out the plan. CCL has been field tested in the past on both a REMUS and a Mid-size Autonomous Research Vehicle (MARV). The newest revision is currently implemented on the SAUV and the TSI modular mission planning toolkit (MMPT) application for glider platform mission planning, monitoring and control.
Distributed Control Environment (DICE) In 2005, the DICE environment was implemented on the SAUV platform as an enabling technology for executing multi-vehicle cooperative behavior. NUWC has developed the Distributed Control Environment (DICE) as a tool for developing behavior-based distributed control systems [8]. It enables communication between distributed system components as well as communication between different systems. DICE has many features specialized for behavior-based systems, thus, it can be useful for development of a wide range of architectures from reactive to deliberative. It supports coordination of processor-intensive tasks, such as high-level planning interacting with responsive low-level control. DICE designers developed this framework to facilitate implementation of multiple autonomous systems that operate with noisy and rangelimited communication, rapidly-changing real-world situations, and variations in resource availability. DICE extends subsumption-style tasking with message passing to the multiagent domain and provides for a wide variety of behavior-arbitration techniques. It allows a great deal of run-time system flexibility including dynamic reconfiguration of behavior structure. DICE is well suited for fast data-driven control strategies. It provides for rapid code development and effective code re-use. Behaviors can be multiply-instantiated and interact through abstract “ports”, which can be dynamically connected to other ports at run-time. Behaviors can be distributed across hosts without code changes.
Cooperative Behaviors The ability of the SAUVs to participate in cooperative role swapping missions has been demonstrated in a “reference mission” as depicted in Figure 2. In general, this mission involves an area which is to be constantly surveyed by a group of SAUVs. In the reference mission, a single SAUV having the highest initial energy takes on the survey role. Those SAUVs not performing the survey maintain a watch circle on the surface recharging their batteries or take on new roles (e.g. mobile communications gateway).
The reference mission builds on the set of behaviors required to implement the mission defined in a related DEPSCoR project entitled “Highly Accurate Temporal and Spatial Mapping of Coastal Regions Using Long Endurance AUVs” (ONR Grant #N000140510666). These include behaviors to support gateway functionality, the survey task, background navigation (including inter-vehicle ranging), networked communications and energy management. These will augment a set of cooperative behaviors already developed, including watch circle, box and lawnmower survey behaviors. These behaviors leverage our CCL and DICE efforts, providing us the ability to implement and test using AUVs in the water.
Networking/Media Access Control (MAC) An underwater Media Access Control (MAC) layer collision handling mechanism which supports ranging as well as communication, while in parallel exploring ad hoc network protocol designs will complement the Autonomous Undersea Systems Network (AUSNET) [9] or Controlled Flooding for Small Networks (COFSNET) [10] designs. Based upon these results, a MAC-layer/network-layer protocol with ability to support system level inputs (e.g. energy, navigation, and mission) and support non-trivial gateway functionality such as packet type queuing and store-and-forward is being developed [11]. This evolved protocol will provide the communications infrastructure necessary to allow platforms to communicate in an ad-hoc, peer-to-peer manner, while supporting the underlying navigation (ranging) requirement and permit platform system inputs to optimize efficiency.
{mosgoogle} Navigation Inter-vehicle ranging will improve underwater dead reckoning (DR) navigation andenable vehicle acoustic ranging. A Time Division Media Access (TDMA) scheme with special “ranging intervals” for each node in the acoustic network has been designed. In this scheme, a node's time slot for transmissions is divided into a communicate interval for normal transmissions and a ranging interval for ranging on all other modems in the network. During its communication interval the source node sends its normal transmissions. During its ranging interval, the source node ranges on all other nodes in the network, collects and stores the results. The resulting information is used by the source node to maintain an understanding of where the other nodes are in relation to it.
Preliminary testing of this scheme took place at Lake George, NY in June, 2006. As part of this effort, collaboration with Teledyne-Benthos, Inc. supports an underwater GPS system [12]. Advanced Operator Planning/Monitoring Tools A collaboration with TSI is developing a Modular Mission Planning Toolkit (MMPT) application to support AUV mission planning. This application provides for planning, monitoring, command and control of multiple heterogeneous AUVs, particularly long endurance glider platforms, with a particular emphasis on incorporating environment data (e.g. currents) into the planning process. Prototype versions of MMPT were used during the 2006 Lake George, NY and Monterey Bay, CA experiments. During the recent AUVFest’07 testing, MMPT was used exclusively to monitor and control a fleet of 3 SAUVs, as well as demonstrate mission planning aspects based on water current METOC data supplied by NRL/Stennis [13].
3. Cooperative Multi-vehicle Field Tests
Two recent major field events involving test and demonstration of multiple vehicles and the technologies previously described are summarized below [14]. Lake George, NY (June 2006) In June of 2006, cooperative behaviors were wet tested in Lake George off Bolton Landing, NY at the RPI Darrin Fresh Water Institute (DFWI). The cooperative mission given to a pair of SAUVs was for one of them to run a box shaped survey while its partner maintained position in a charging mode. The cooperative mission statement specified only that the vehicle with the most energy was to run the survey and the other was to charge. The vehicles were to decide which one would take on the survey role at runtime. The successful experiment demonstrated several autonomous role switches based on the vehicles’ relative energy levels. This indicated that the initial design of the high level behavior logic as proven under simulation conditions transitioned well into the real world. It also demonstrated the utility of the COFSNET underwater networking protocol for three nodes (2 SAUVs and a gateway buoy/operator).
Figure 3 shows a screen shot taken during that experiment. The entire image shows the
Macromedia Breeze system in use, where the output of various applications appears in Breeze sub-windows. Shown to the right is an early prototype MMPT application plotting the vehicle positions along with data pulled out of the vehicles’ status messages. The column of sub-windows on the left shows an ongoing live chat-like interaction between MCAUV team members at Bolton Landing, NY, Wiscasset, Me (TSI), and Newport, RI (NUWC) as well as photos taken on the lake. Using this system, collaboration of remote participants with onsite operators in an on-going multi-vehicle field event was demonstrated.
4. Cooperative Multi-Vehicle Navigation Algorithms [23] In this project, strategies that integrate the capabilities of the SAUV as a GPS node while on the surface with inter-vehicle range sensing based on acoustic ranging to achieve improved localization and navigation capability of the group are being studied. A model for dead reckoning provides estimates for the expected performance of the single SAUV under general conditions and control missions. The mathematical tools of modeling and data fusion are general, and can be applied to any possible trajectory and mission. In one example, a dead reckoning model for vehicle v1 yields a performance of 5% error predicting a localization of 50 meters after 1 kilometer. Errors of this magnitude are commonly observed in SAUV operations tests. The multi-vehicle sensor fusion strategies described below are intended to dramatically improve this performance.
In multivehicle operation, a second SAUV vehicle v2 on the surface may provide a GPS reference measurement for the system, as depicted in Figure 1. In this two-dimensional scenario, an acoustic link is established between v1 (submerged) and v2 (surface). This acoustic link is used to measure the range R12 between v1 and v2. A number of alternative technologies exist for such acoustic range measurements and new, more accurate, approaches are being explored.
An additional strategy further improves this performance by taking advantage of the ability of the GPS-SAUV on the surface to move, tracking the motion of v1. As shown schematically in Figure 2, v2 may move (continuously, or intermittently) on the surface in order to reduce the range itself, and therefore reduce the associated range errors.
The strategies of sensor fusion and tracking described above illustrate the role of two cooperating vehicles moving in formation to achieve improved localization and navigation. In this simplified example, only x-direction localization has been considered. In order to also account for lateral localization (y-direction) additional vehicles must be added. Two multi-vehicle configuration strategies are described below that provide both x-direction and y-direction localization.
Figure 3 illustrates the strategy MV3A, in which two SAUV’s, v2 and v3, are positioned on the surface with GPS localization. A single vehicle, v1, is submerged and range measurements are obtained with respect to both v2 and v3. Again, depth measurements are provide information to triangulate the position of v1. In the MV3A scenario, the Kalman filter provides data fusion of two range measurements with dead reckoning, and determines estimates of the components of the covariance matrix.
Figure 4 illustrates the strategy MV4B, in which four SAUV’s are employed. One SAUV, v2, is on the surface, and three are submerged. The analysis in this case assumes an initial reference configuration for v3 and v4, and can be obtained from directional velocity measurements of v3 and v4 relative to v2 (at known depth, d and known compass directions. Given this reference configuration, the Kalman filter data fusion proceeds among the four vehicles (six range measurements are made and fused).
References [1] D. R. Blidberg and D. M. Crimmins, “Multiple Cooperating AUV (MCAUV) Digest,” Naval Undersea Warfare Center Division Newport, Newport, RI, December 2005. This work was performed under ONR Grant Number N00014-04-1-0264 and was completed March 2006.
[2] J. Jalbert, et al., “Solar-Powered Autonomous Underwater Vehicle Development,” in Proceedings of the Thirteenth International Symposium on Unmanned Untethered Submersible Technology, August 2003.
[3] S. G. Chappell, S. S. Mupparapu, R. J. Komerska, and D. R. Blidberg, “SAUV II High Level Software Architecture,” in Proceedings of the Fourteenth International Symposium on Unmanned Untethered Submersible Technology, August 2005.
[4] D. E. Humphreys and R. D. Schulz, “A Hydrodynamic Analysis of the Performance of the Falmouth SAUV,” Vehicle Control Technologies, Inc., VCT Tech Memo 06-11, October 2006.
[5] R. J. Komerska and S. G. Chappell, “A Simulation Environment for Testing and Evaluating Multiple Cooperating Solar-powered AUVs,” in Proceedings of the MTS/IEEE Oceans 2006 Conference, September 2006.
[6] C. N. Duarte, et al., “An Update on a Common Control Language to Support Multiple Cooperating AUVs,” in Proceedings of the 15th Annual International Symposium on Unmanned Untethered Submersible Technology (UUST07), August 2007.
[7] D. R. Blidberg, “Generic Behaviors: Definition and Structure, an Approach to Modularity,” in Intelligent System Control Architectures – Volume 1: Technical Proposal, Proposal for SOL BAA #94-19, July 21, 1994.
[8] C. N. Duarte, et al., “A Common Control Language to support Multiple Cooperating AUVs,” in Proceedings of the 14th Annual International Symposium on Unmanned Untethered Submersible Technology (UUST05), August 2005.
[9] S. S. Mupparapu, R. Bartoš, and M. Haag, “Performance Evaluation of Ad Hoc Protocols for Underwater Networks,” in Proceedings of the 14th International Symposium on Unmanned Untethered Submersible Technology, August 2005.
[10] R. Bartoš, R. Komerska, S. G. Chappell, and M. Haag, “COFSNET+: A Routing Protocol for Communication among Autonomous Underwater Vehicles,” in Proceedings of the 15th Annual International Symposium on Unmanned Untethered Submersible Technology (UUST07), August 2007.
[11] M. Haag, E. Agu, R. Komerska, S. G. Chappell, and R. Bartoš, “Status Packet Deprecation and Store-Forward Routing in AUSNet,” in Proceedings of the First ACM International Workshop on UnderWater Networks (WUWNet), September 2006.
[12] J. Jalbert, et al., “SAUV II Operations and Technology Transfer into Autonomous Station-Keeping Gateway Buoys,” in Proceedings of the 15th Annual International Symposium on Unmanned Untethered Submersible Technology (UUST07), August 2007.
[13] R. W. Nitzel, M. Haag, C. Yohman, C. Benton, R. J. Komerska, and S. G. Chappell, “An Update on the Development of the Modular Mission Planning Toolkit (MMPT),” in Proceedings of the 15th Annual International Symposium on Unmanned Untethered Submersible Technology (UUST07), August 2007.
[14] S. G. Chappell, et al., “Recent Field Experience with Multiple Cooperating Solar- Powered AUVs,” in Proceedings of the 15th Annual International Symposium on Unmanned Untethered Submersible Technology (UUST07), August 2007.
[15] D. O. Popa, A. C. Sanderson, R. J. Komerska, S. S. Mupparapu, D. R. Blidberg, S. G. Chappell, “Adaptive Sampling Algorithms for Multiple Autonomous Underwater Vehicles,” in IEEE/OES AUV2004: A Workshop on Multiple Autonomous Underwater Vehicle Operations, June 2004.
[16] Sanderson, A.C., Vadiraj Hombal, David P. Fries, Heather A. Broadbent, James A. Wilson, Pragnesh I.Bhanushali, Stanislav Z. Ivanov, Mark Luther, Steve Meyers, “Distributed Environmental Sensor Network: Design and Experiments”, Proceedings of the 2006 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems, Heidelberg, Germany, September 3-6, 2006.
[17] Hombal, V., A. C. Sanderson, and D. R. Blidberg, “A Non-Parametric Iterative Algorithm for Adaptive Sampling and Robotic Vehicle Path Planning”, Proceedings 2006 IEEE International Conference on Intelligent Robots and Systems, Beijing, China, October 9-13, 2006.
[18] Sanderson, A.C, and D. R. Blidberg, “Distributed Sensor Networks and Environmental Robotics”, Proceedings of the IAGLAR Annual Conference on Great Lakes Research, May 20-26, 2006.
[19] Xue, F., A.C.Sanderson, and R. Graves, “Multi-Objective Routing in Wireless Sensor Networks with a Differential Evolution Algorithm”, Proceedings 2006 IEEE International Conference on Networking, Sensing, and Control, April 23-25, 2006, Ft. Lauderdale, Florida. 880-886.
[20] Fries, D., H. Broadbent, J. Wilson, P. Bhaushali, S. Ivanov, and A. C. Sanderson, Aquatic Microsystems Technology Sensor Networks For Environmental Monitoring Utilizing Optimized Sampling Algorithms”, Proceedings 2006 AGU Ocean Sciences Conference, Honolulu, HI, February 20-24, 2006.
[21] Chappel, S., A.C.Sanderson, and D.R.Blidberg, “Multiple Cooperating Vehicles (MCCAUV) Research Program”, Proceedings of ONR Conference on Unmanned Systems Technology Development, Panama City, FL, February 6-10, 2006.
[22] S. Mupparapu, S. Chappell, R. Komerska, and Richard Blidberg, “Autonomous Monitoring and Control (ASMAC) – An AUV Fleet Controller,” in IEEE/OES AUV2004: A Workshop on Multiple Autonomous Underwater Vehicle Operations, June 2004.
[23] D. Mirabello, A. C. Sanderson, and D. R. Blidberg, “Comparing Kalman and Particle Filter Approaches to Coordinated Multi-Vehicle Navigation,” Proceedings of the 15th Annual International Symposium on Unmanned Untethered Submersible Technology (UUST07), August 2007.
Christian Louboutin shoes are very popular among fashionable ladies and young girls, frot its character of elegance and noblity. So do the Herge Leger Dresses. This is why many girls like wearing the shoes going with the dress. Many young ladies are accustomed to dress themselves sitting in front of the mirror and decide whether they should wear Links of London or American stylish Tiffany Jewelry. When they fournd it is snowing outside and they will surely want to protect their feet to wear UGG Boots and select MBT Shoes for their husbands who are watching some attractive films from Wholesale DVD shelf. Referring to the gender difference, brave gentlemen will take the
Vibram Five Fingers Shoes with them and explore a chance for adventure. Life is so colorful with all sorts of products we can offter, do you agree with me?
