The Evolution of Mine Countermeasures Autonomy at Panama City (Web Exclusive)

Autonomous Mine Countermeasures is a capability now in the fleet that was a long time in coming--an effort made possible by researchers at the Naval Surface Warfare Center Panama City Division. (Photo by MC1 Gary Keen)

Autonomous mine countermeasures–such as this REMUS–is a capability now in the fleet that was a long time in coming. It was an effort made possible in part by researchers at the Naval Surface Warfare Center Panama City Division. (Photo by MC1 Gary Keen)

By Matthew Bays, Signe Redfield, John Hyland, Jim Perkins, Cheryl Smith, and Gregory Garcia

The mine countermeasures (MCM) mission has long been a prime candidate for autonomous solutions. MCM operations are repetitive and time consuming, and they put personnel in a high-risk environment where mistakes can have catastrophic consequences. Aspects of the undersea domain, such as communications constraints, also lend themselves to the development of autonomy. As the Navy’s technical center of excellence for littoral warfare and coastal defense, the Naval Surface Warfare Center Panama City Division has advanced the application of autonomy as an effective way to support MCM missions. Early autonomy research established baseline capabilities and hardware functionality to perform simple tasks. Today’s research focuses on the algorithms, adaptations, and evaluations required in the MCM mission area. The research into MCM autonomy is by no means exhaustive in our outline of where it has been, where it is, and where we hope it will go. Indeed, there are numerous other groups in academia and industry working on MCM autonomy. Rather, we wish to share its evolution through the lens of how it has changed throughout the years at Panama City.

Autonomy from Then to Now
As unmanned systems were developed for time-limited MCM surveys, the need for autonomy became increasingly evident. Untethered, unmanned underwater vehicles were seen as the future of mine hunting because of their relatively low cost, unmanned nature, and ability to get close to mines. Once the tether connecting an unmanned underwater vehicle to its operator is cut, however, the burden of transmitting information acoustically to the operator becomes unmanageable. Acoustic communication offers orders of magnitude less bandwidth than in-air communication, only allows one vehicle to communicate at a time, and is prone to data loss and error. These limitations drive the need for underwater vehicles to perform autonomously.

In the earliest days of research into MCM autonomy and unmanned systems, the most difficult challenges were the most fundamental: creating a functional platform, enabling it to localize itself within its environment, and enabling it to follow a list of predefined waypoints. Early examples of this focus on fundamentals at Panama City include the Remote Undersea Mine Countermeasure (RUMIC) system and the Very Shallow Water Autonomous Underwater Vehicle (VSWAUV). RUMIC was a modified MK-7 swimmer delivery vehicle with a side-scan sonar, navigation instrument suite, and computer to control the actuators and propulsion system. It was one of the first applications of Doppler-inertial navigation technology. VSWAUV, based on a MK-37 torpedo hull, was intended to show the feasibility of autonomous mine search in very shallow water. Both vehicles were capable of executing preprogrammed search tracks with a few self-preservation actions, such as monitoring depth to avoid diving too deep and monitoring altitude above bottom to avoid grounding.

Early investments produced a variety of solutions to the most essential problems, including localization, navigation, and basic vehicle safety. Modern researchers are now tackling more challenging problems. Yesterday’s “mission autonomy,” the autonomy research directly related to the execution of a specific mission, has become today’s “infrastructure autonomy,” the established tools providing basic non-mission-specific functionality. Today’s research determines the fewest number of waypoints in a vehicles path with least risk in the event of limited resources, what order in which to prosecute waypoints, or even how to allocate waypoints among multiple vehicles.

As research on unmanned underwater vehicle platform autonomy has advanced, Panama City’s focus has broadened to include developing the requirements to transition these systems and investigate the tactical implications of their use. In the 1990s, Panama City developed the Autonomous Littoral Warfare Systems Evaluator, which incorporates both a low-fidelity Monte Carlo simulation tool (a research method based on repeated random sampling) for evaluating autonomous behavior, and a high-fidelity, engineering-level simulation tool that allows detailed characterization of environmental effects, vehicle dynamics, and other physics models into a testing environment for autonomous vehicles. Current research includes mission autonomy, as well as a reevaluation of established MCM concepts of operation and more sophisticated test and evaluation tools and methodologies to maximize the benefit of autonomy.

Mission autonomy research is focused on developing capabilities that enable vehicles to accomplish specific missions, from surveys to reacquisition and identification of targets to neutralization. Current efforts include more efficient algorithms for reacquiring and identifying targets, adaptation to the environment for more effective surveys, and decision-making mechanisms enabling multiple vehicles to determine the best way to allocate these tasks to individual vehicles. Before these decision-making mechanisms can be used, however, the individual vehicles must be capable of accomplishing their tasks.

Single-Vehicle Mission Autonomy
Multiple projects are underway to provide individual vehicles with the means to adapt and improve their contributions to the MCM mission. Under the Office of Naval Research’s integrated forward-looking sonar and synthetic-aperture side-look sonar program, Panama City has developed a modular, open architecture that facilitates flexible development and rapid reconfiguration and interfaces with the project’s advanced sensors. Combined with appropriate behaviors, vehicles using this architecture can perform large-area, single-pass preparation-of-the-operating-environment surveys while avoiding obstacles.

This architecture uses a three-layered approach consisting of a high-level mission manager, an intermediate layer behavior engine, and a low-level path planner. This approach balances the deliberative, long-term decision making needed to achieve goals with the reactive, short-term capabilities needed to operate in uncertain and dynamic environments. The architecture’s layered interfaces facilitate the addition of new system tasks and tactical behaviors with minimal effects to existing software components. Existing architectures often fail to make the decision-making component modular, limiting future enhancements, but this design supports the program’s incremental technology development strategy. It prioritizes the development of autonomy for mission safety first, then mission execution, and finally mission optimization using environmental and tactical knowledge.

One example of such modular behaviors is the warfare center’s multiple-aspect coverage algorithm, which plans paths for reacquisition and identification tasks. Within each task, vehicles must revisit targets with a high-resolution, low-coverage sensor and determine whether each is a target of interest. The algorithm breaks a target list into clusters and creates vehicle paths to inspect the objects from multiple directions within each cluster. Instead of inspecting each target individually, the algorithm develops customized inspection patterns that examine multiple targets at once. The algorithm uses formal probability theory to determine patterns that meet desired probabilities of reacquiring and identifying every target. These patterns can reduce the time required to reacquire and identify a group of targets by as much as 40 percent in high-clutter environments.

Multivehicle Mission Autonomy
In addition to research enabling individual vehicles to become more capable, multivehicle efforts also are underway. Multivehicle approaches can significantly improve mission performance by dividing one large task between resources or handling multiple small tasks in parallel, but they require communications to achieve maximal benefit. When multiple vehicles can operate in the same area, with onboard obstacle avoidance and deconfliction, they can share the load of a joint search mission more effectively and respond quickly to new tasks that arise during the search. For instance, multiple vehicles performing a collaborative search may identify a potential target for inspection. The Wide-Area Search project enables researchers to experiment with modular decision-making mechanisms (arbiters). Physical and virtual underwater assets autonomously performed joint surveys and reacquired virtual targets, while a physical surface asset received target information and scheduled neutralization tasks. The neutralization was simulated with a tele-operated remotely operated vehicle.

Current MCM planning for multiple assets involves scheduling tasks independently of one another. New tasks are expected to arise from the data gathered during the current tasks, but those new tasks are viewed by the scheduler as unexpected interruptions, preventing efficient asset scheduling. The new Optimized Waterspace Management project will incorporate required mission constraints, asset requirements, and expected additional information to sequence the tasks in an optimal manner.

Tactical Shifts
Progress is being made toward improved capability against specific mission elements. As these systems come closer to transition, however, additional roadblocks to their acceptance become apparent. Not all of these are engineering or cultural problems—several are related to a lack of existing tools to support the analyses of these systems’ performance. Panama City is actively engaged in the research required to develop these tools, which includes shifting the tactical paradigm for working with autonomous systems and determining how best to evaluate new elements of mission autonomy.

When you substitute autonomy for automation without providing mission decision-making opportunities onboard, you lose potential performance gains from the autonomy. Current tactical paradigms constrain autonomous systems to automated roles such as following a path predefined by an operator and returning home. This constraint imposes artificial restrictions on improvements to operator workloads and overall effectiveness. Revised tactical assumptions will be necessary to use autonomous systems most effectively.

For example, Panama City has been developing automated perception systems for decades. Automated target recognition tools, initially envisioned as operator aids, are being improved and adapted to provide sensory perception for underwater vehicles while underway. Vehicles that can use this sensor data to change their actions could improve performance in unexpected environments. Currently, plans are generated against anticipated environments—not the actual environment. Subtle differences in environment can have a great effect on sensor and mission performance. An accurate understanding of the environment enables a vehicle to plan and execute an improved mission. Discriminating between potential targets and the environment enables vehicles to decide whether to use the mission time available for a closer look at a target, or for a larger initial search, increasing the efficiency of limited in-water endurance and reducing the burden on human operators. Under current tactical paradigm, there has been no formal mechanism for a vehicle to adapt its mission based on information it collects.

Test and Evaluation
If MCM autonomy tools in development are to transition to the future force, we need ways to judge their effectiveness as they make their way through acquisition to warfighters.

Two autonomy test bed technologies are under development: a 3-D virtual reality mission viewer and a PixelSense (Surface Table) multiuser, multivehicle, intuitive mission planner. At the lowest level of test and evaluation, the goal of the project is to provide hardware-in-the-loop simulation capabilities for virtually any autonomous underwater vehicle. This would create a low-cost alternative to at-sea testing used to develop and test autonomous behaviors, concepts of use, and multivehicle collaboration. This simulation capability permits greater opportunity to experiment with unmanned systems, enabling real-time observation of vehicle decisions and responses for troubleshooting and quantifying vehicle performance in a variety of “what-if” scenarios.

Addressing mid-level evaluation for surveys, the System Performance and Layered Analysis Tool—originally developed for timely detection of terrorist threats in open, crowded areas—is being modified for use in autonomous mine warfare and MCM. Given a configuration of multiple stationary threat detection sensors and a scenario specification, it detects moving targets. This circumvents the need to model complex sensor performance functions mathematically and enables it to provide timely, accurate estimates of a moving target’s location based on the given sensor locations. This work is the first step toward a reconfigurable sensor system, where assets determine the optimal positions for detecting targets given changes in the environment.

At the highest level, the MCM mission must integrate with battlespace strategy and timeline. In conjunction with university and industry researchers, Panama City is developing a simulation tool to help researchers evaluate the impact of autonomous solutions in larger systems. Existing measures of effectiveness are based on performance evaluations after missions are complete. Autonomous systems that can adapt their tactics require a new kind of metric—a run-time measure of effectiveness that can be evaluated during the mission.

Conclusions
MCM autonomy research has come a long way in the past decade. The focus is no longer on developing custom-made research platforms that have basic functionality, but rather on providing advanced MCM mission-centric capabilities to commercial platforms, including adaptation, cooperation, and optimization. In addition, researchers are finding how best to change the tactical paradigm from one focused on operators performing the majority of the MCM labor to one that leverages the unmanned system’s autonomy to accomplish the high-level goal of clearing the minefield.

About the Authors:
Drs. Bays, Hyland, Smith, and Garcia are autonomy researchers at the Naval Surface Warfare Center Panama City Division. Dr. Redfield is an autonomy researcher at the Naval Research Laboratory. Jim Perkins is a computer scientist at NSWC PCD focusing on autonomy research.