Feature ArticleCollaborative Autonomous Vehicle Use in Mine Countermeasures
By Dr. Vladimir Djapic
Scientist, Task Leader
Autonomous Naval Mine Countermeasures Program
Systems Technology Department
NATO Undersea Research Centre
La Spezia, Italy
Faculty of Electrical Engineering and Computing
University of Zagreb
Autonomous underwater vehicles (AUVs) and autonomous surface vehicles (ASVs) used by the operational community are currently limited to executing preplanned missions. Human operators must intervene to observe events, make decisions and guide the vehicle, and implementing a data-reactive plan requires working with vehicle manufacturers to develop a custom solution.
For mine countermeasures (MCM), creating a single autonomous platform with a data-reactive mission planner may be prohibitive in terms of time, cost or effort, and operators may be unfamiliar with such autonomous vehicle planning, guidance and control.
Collaborative autonomous vehicles have the potential to lessen the burden by conducting autonomous MCM missions. These vehicles could transform MCM capabilities from legacy systems deployed from surface ships (50 sailors working in a minefield) to a quickly deployable system. Collaborative autonomous vehicles are designed to be scalable to the needs of the operation and to offer an order of magnitude increase in the tempo of operations, which comes with a reduction in through-life costs by keeping personnel out of mine fields. The demonstrated benefits of autonomous behaviors involve minimizing the operator workload associated with autonomous platforms.
Research being conducted by the NATO Undersea Research Centre (NURC) and the University of Zagreb strives to design autonomous systems with data-reactive mission planning capabilities utilizing open-architecture principles for software and hardware.
NURC is pursuing such a system using a squad of autonomous vehicles with complementary capabilities that use cooperative behavior to most effectively achieve a mission. The simplest squad consists of two vehicles: An ASV and AUV squad can conduct the mine detection, classification, localization and identification portion, and an ASV (or AUV) and unmanned underwater vehicle (UUV) can conduct the mine reacquisition and neutralization portion.
Current efforts focus on cost-effective neutralization (end-state driver) using a pair of autonomous platforms: a highly capable ASV (it could also be a highly capable AUV) and a guided, low-cost neutralization UUV. The result of this research is the development of a prototype autonomous naval mine countermeasures (ANMCM) system. An AUV and ASV squad of this system was recently demonstrated in NURC's CATHARSIS II sea trials, which were conducted in October 2009.
Difficulties With Autonomy
In surveillance and reconnaissance of objects of interest in the sea, sophisticated methods have been developed for detection, classification, localization and identification.
However, performing these functions reliably on objects embedded in complex, transient, cluttered environments has proven to be challenging. False alarm rates remain stubbornly high, which can be troublesome if the end-state is intervention and if all intervention options are costly or irreversible.
Mine reacquisition and neutralization. A highly capable SWATH ASV or hover-capable AUV guides a low-cost expendable UUV to the sea mine.
If the process of detection through neutralization is treated as a fully coupled integrated system, then a number of favorable trade-offs emerge from creating new intervention options. Low-cost and nonlethal intervention methods may considerably relax accuracy demands and provide flexibility for the responder. For example, an electronic mine neutralization device can be used as the payload instead of an explosive device, in which case the UUV might not be destroyed.
Concept of Operations
Mine neutralization relies on accurate detection, classification, localization and identification. AUVs with synthetic aperture sonar (SAS) sensors are seen as the first part of a collaborative chain for autonomous MCM. The goal is to detect and classify the mine with super-classification SAS sonar (a sonar with the ability to recognize mines at long range and with significant confidence). An example of such a SAS AUV system is the MUSCLE AUV developed at NURC. This vehicle could be delivered to the operational area by a highly capable ASV; for example, a small waterplane twin hull (SWATH) ASV.
In the envisioned concept of operations, a highly capable autonomous vehicle (an ASV or AUV) would reacquire a previously identified target using its imaging sonar. The vehicle then initiates an adaptive engagement plan autonomously (it may also be controlled by an operator if the communication link exists). The vehicle tracks, scans and maintains the target in the field of view of the imaging sonar, while compensating for wind, waves and currents. Next, the autonomous vehicle guides the low-cost, simplified UUV to deploy a payload to neutralize the target. Finally, the vehicle performs a neutralization assessment to determine and report the degree to which the mission has been accomplished.
Collaboration between the highly capable autonomous vehicle and the low-cost mine-neutralization UUV is viewed as an essential element of reliable mine neutralization. UUV guidance is accomplished by automatically processing sonar imagery and passing sonar position, range and bearing information to the UUV, which processes this information to drive itself to the desired location.
Autonomous MCM Technologies
For the prototype ANMCM system, researchers have designed a system consisting of three vehicles: an AUV, an ASV and a UUV.
AUV Prototype. The MUSCLE SAS AUV uses a navigation and control system built by Bluefin Robotics Corp. (Cambridge, Massachusetts) and a SAS sensor from Thales Underwater Systems (Brest, France). For surface navigation, the AUV uses a global positioning system (GPS) sensor. For underwater navigation, the AUV uses a combination of a Doppler velocity log and an inertial measurement unit.
ASV Prototype. An ASV transports the AUV to the operational area, conducts tracking while the AUV is submerged and delivers the neutralization UUV. The ASV would be equipped with an imaging sonar, an ultra-short baseline sensor and navigation sensors similar to the AUV. The NURC prototype ASVs are rigid hull inflatable boats (RHIBs) powered by an outboard engine made by H-Scientific Ltd. (Waterlooville, England) and a catamaran type of ASV made by SeaRobotics Corp. (Palm Beach Gardens, Florida). Future plans are to demonstrate external system guidance technology on a SWATH-type ASV and/or hover-capable AUV. The ASVs carry BlueView Technologies Inc. (Seattle, Washington) forward-looking imaging sonar operating at 450 and 900 kilohertz.
UUV Prototype. An expendable neutralization UUV utilizing commercial-off-the-shelf components, a VideoRay (Phoenixville, Pennsylvania) Pro4 remotely operated vehicle (ROV), was designed. Batteries, an acoustic link (umbilical removed) and automation navigation and control algorithms were added to this ROV. A nonexplosive electronic neutralization device was developed and will be miniaturized to serve as a potential payload for the UUV prototype.
Autonomy Software Backbone
The common glue that would link the platforms and enable multivehicle collaboration is the autonomy software backbone. Although there are various software architectures that allow for behavior-based planning and control, the architecture chosen was Mission Oriented Operating Suite Interval Programming (MOOS-IvP) autonomy architecture.
MOOS-IvP is a suite of open-source software facilitating interprocess and interprocessor communication via a publish and subscribe paradigm. NURC researchers implemented a nonlinear ASV controller based on backstepping as a backseat driver, which outputs desired signals for more precise tracking of complex trajectories. While low-level control tasks such as navigation, depth keeping and vehicle safety are assigned to the ASV main-vehicle computer (front-seat driver), all high-level control inputs are derived from a separate vehicle payload computer running the MOOS system (backseat driver).
Guidance and sonar processing is performed on the backseat computer. The algorithms are generic and can be ported to any future platform.
Access to actuators from the backseat computer is not allowed. Course, course rate and speed control are performed by the front-seat computer. MOOS processes communicate through a MOOS database in a publish-subscribe manner. Variables of interest are published to this database by processes, while others subscribe to variables they need.
MOOS is also used in the prototype system to pass tracks from the imaging sonar. Output from a sonar sensor is used to direct the ASV (or possibly the future AUV) to change its trajectory as new mission plans are developed on board the vehicle in response to the sensor data.
Automated tetherless UUV tracking (white dot) from the sonar imagery.
Field Test and Future Plans
During the CATHARSIS II sea trial offshore Elba, Italy, a conceptual demonstration of the mine neutralization phase was demonstrated using an ASV equipped with short-range, high-resolution sonar systems and a high-end navigation system.
It was demonstrated that the RHIB ASV can reacquire a bottom target with imaging sonar, maintain position while pointing at the target and perform a circular search around the target while varying the circle diameter.
A lawn-mower pattern search was initiated around the point previously determined to be a mine-like target location. Upon target detection and click-to-contact operator intervention, the ASV initiated the appropriate behavior.
During the sea trial, the group successfully tested the interaction between the operator, the autonomy software and the ASV. It allowed the operator to click on an interesting object on the display showing the sonar data, which prompted the vessels to plan a course to retrace and investigate the object. The location of the point is calculated by knowing two things: the time of the ping and the distance abeam of the vessel (port or starboard).
When combined with GPS information and the heading information from the vessel, an absolute location of the click can be calculated. The ASV promptly changes course to drive while maneuvering relative to the bottom target.
In the future, this lower resolution sonar feedback can be used for guiding the simplified mine neutralization weapon to moored or bottom mines, as the mine would have already been detected and classified during the prior super-classification mission. As automatic sonar image processing algorithms mature, rather than the operator making the decision, autonomous vehicles could be able to make a decision on target detection during the target reacquisition phase.
During the Autonomous Neutralization Trial offshore Elba this month, a catamaran-type ASV—equipped with variable depth and pan-and-tilt sonar capabilities, along with external guiding algorithms—will acoustically guide an inexpensive neutralization UUV to the bottom target.
This paper describes results that show potential for collaborative use of autonomous vehicles in MCM. A common goal has been using autonomous vehicles to minimize operator workload. NURC's prototype auto'no'mous systems are designed to be scalable to the needs of the operation and to offer an increase in the tempo of operations and reduction of life-cycle costs.
The critical task of positioning the vehicle such that the classification sensors can be effectively brought to bear was the primary goal of the group's autonomy development.
Future work will expand the role of the autonomous platform to allow deployment of MCM neutralization devices. By testing the developed algorithms on board ASVs, it will be possible to gather the necessary data to determine the most appropriate reacquisition platform for various scenarios encountered in MCM.
The importance of the use of autonomous unmanned systems here extends beyond the benefit of reduced operator workload and shows a clear path for increased operator safety.
The authors thank Dr. Marc Pinto for sonar-based guiding ideas and Dr. Tom Curtin for financial support. The ASV experiments reported here owe their success to the computer programmers, engineers and technicians who integrated new functionality into the vehicles and kept them operational in sometimes difficult environments, especially Dr. Stefano Fioravanti, Tom Pastore, Arjan Vermeij and Alberto Grati. The authors thank BlueView Technologies for loaning a sonar unit for the purpose of this experiment and also thank Henry Robinson from H-Scientific Ltd. for support.
Dr. Vladimir Djapic joined the NATO Undersea Research Centre as a scientist in the systems technology department of the Autonomous Naval Mine Countermeasures program in 2008. His research interests include applying behavior-based, nonlinear and adaptive algorithms to autonomous control of unmanned vehicles. The objective of his research at NURC is to design a low-cost, robust and effective mine neutralization system.
Ðula Nađ is a Ph.D. candidate in control engineering at the Faculty of Electrical Engineering and Computing, University of Zagreb. In 2009 he participated in the visiting researcher program at the NATO Undersea Research Centre in La Spezia, Italy. His research interests include control of autonomous vehicles, underwater acoustic navigation and nonlinear systems.