Software System Manages Schedules for Troops

October 2006
By Rita Boland

Coordinators is an artificial intelligence system that can manage scheduling for warfighters, freeing them to focus on other mission aspects. The software keeps track of all the tasks and assignments for individuals or units. If three units were poised for a synchronized strike and one unit was delayed, its Coordinator could contact the other units’ Coordinators and resynchronize the mission.
Artificial intelligence program would allow warfighters to focus less on coordination and more on objectives.

Troops in the field may soon have a little help handling their busy schedules. Researchers are working on a developmental distributed intelligent software system that adapts field units’ mission plans as situations and events unfold. The software can be used with a variety of devices and reduces the time and personnel necessary when changing tactics.

Three teams—the University of Southern California’s Information Sciences Institute, Marina del Rey, California; Honeywell International Incorporated, Morristown, New Jersey; and SRI International, Menlo Park, California—are developing the software, called Coordinators. The Defense Advanced Research Projects Agency (DARPA), Arlington, Virginia, serves as the lead on the Coordinators project and sponsors the teams.

“The teams are creating complementary Coordinator solutions that have different strengths,” explains Dr. Tom Wagner, DARPA program manager. “For instance, one technology will be particularly well-suited to applications in which communications are very limited, while another may perform better in situations when changes occur very frequently.

“The teams are also creating complementary technologies that accompany the coordination core,” Wagner says. “For instance, one team is creating a module that learns soldier decision preferences so that Coordinators always generates the best options first in time-constrained situations, while a different team is learning soldier response times so that Coordinators can plan for the time required to obtain human input.”

Coordinators keeps track of all the tasks and assignments for individuals or units. Each individual scheduling element is referred to as a Coordinator and together work as Coordinators. Wagner gives the example of three units, each with its own Coordinator equipped with the mission plan, poised for a synchronized strike. If one unit was delayed, its Coordinator could contact the other units’ Coordinators and resynchronize the mission.

“The goal of the program is to create systems that provide intelligent coordination decision support to soldiers so that when changes occur, they can select from options generated by the Coordinators rather than manually exchanging information and manually adapting plans and tasks,” Wagner states.

Coordinators can reason about mission tasks, timing and changes. The system also can quickly determine how one change will affect other events and can identify the best options or choose from a selection of preplanned contingencies. Currently, battle plan adaptation is manual. The software would cut down on the time commanders need to rearrange battle plans, reducing human error, he adds.

For very large operations, changes to a battle plan can affect the schedules of numerous units. In some scenarios, Coordinators may link the mission plans of hundreds or thousands of troops. The software could manage changes quickly and alert all users to the adjustments.

“With this new technology, units will respond to change more rapidly, more accurately, with less cognitive load and will exhibit a greater degree of coordinated action,” Wagner emphasizes. “We believe Coordinators will save lives and make our forces even more effective.”

Whether the Coordinators software is being used by just a few individuals or by several hundred, these personnel would have access to the mission requirements recorded in the software, explains Dr. Charlie Ortiz, program director of the Teambotics program within the Artificial Intelligence Center at SRI. “You’ve got each of these Coordinators, and at time zero before the mission starts, they’d all be loaded up with the mission,” he states.

The Coordinators software is not hardware-specific and can be loaded onto devices such as personal digital assistants. An individual Coordinator functions as a node in the overall software package. In a network of 100 nodes, each is a scheduling element corresponding to a certain person. Coordinators can make changes to the schedule independently or after asking for user consent.

For example, if a two-hour adjustment to the schedule occurs, Coordinators would notify personnel about the alteration. Because each Coordinator is programmed with its user’s schedule, it also knows when not to interrupt for safety or other reasons. If a unit were taking a bridge at the time of a schedule change, the Coordinator would not interrupt the troops at that time. While a Coordinator would make that decision based on the schedule, eventually there may be a more intelligent way for the software to make its choices.

“There’s research that’s going on in an area called multi-agent systems that models the way agents or intelligent entities such as humans and software programs can negotiate on how to make tradeoffs with responsibilities,” Ortiz says.

Changes to the schedule can be made in several ways. Sensors can be integrated into the software to make adjustments automatically when certain events occur. Or, individuals can enter new information that affects plans. Depending on how the system interfaces with other systems, information can be entered through different methods. At minimum, individual users will always be able to make changes.

Regardless of who or what makes the schedule alteration, Coordinators works to manage every objective affected by the change. If an individual knows a task has to be completed at a certain time, Coordinators will try to work it into that person’s schedule. When that is not possible, the system will search through its other elements to find an individual or unit that can complete the task. When alterations are made, the software can devise the necessary shifts of responsibility.

The software’s ability to determine what questions to ask and when to ask them is called coordination autonomy. This allows an individual Coordinator to tailor itself to its user’s schedule and to know when it should alert the user or make independent adjustments. “The system should learn about how its user interacts with it to tailor that process in the most unobtrusive fashion,” Ortiz says.

In the near term, Coordinators will provide only decision support and will not make autonomous choices. However, developers envision Coordinators learning the amount of time an individual takes to respond and learning a person’s reaction to interruptions when performing a task.

“This means that the system can be very sophisticated about interacting with the soldier,” Wagner explains. “That is, it will know how certain it is about a given decision, how critical the decision is, how long it can wait for a response from the soldier, how long the soldier will take to respond and so on. From this information, it can decide whether to ask the soldier, wait to ask or decide autonomously—if the system were very certain—if the time constraints require it.”

Timing difficulties are not the only problems Coordinators will have to manage in the field. The system will need to handle issues such as constricted or constrained communications. “Everyone is not always connected in the real world because things can go wrong,” Ortiz states. One of the benefits of the software, according to its developers, is its ability to function even when some nodes cease to work properly. If communication with one node fails, the program continues to operate. Also, if a node is eliminated, the system takes that into account and reallocates the responsibilities. In addition, if a new unit or individual is added, its node is integrated into the schedule and responsibilities are shifted.

Though Coordinators can process numerous changes, Ortiz recommends keeping schedules as stable as possible for the benefit of users. “I probably won’t become very good friends with you if my Coordinator keeps asking you to modify your schedule every two minutes,” he says.

While stable schedules are best, the Coordinators software gives troops the ability to react quickly to unexpected events in a dynamic battlefield. “It’s a centralized decision maker,” Ortiz explains.

Coordinators developers emphasize that the software is a new way for scheduling dispersed operations. “This is the first intelligent system for dynamic distributed military task coordination,” Wagner says.

The thrust behind the development of the technology is providing a relevant tool to the troops. The system frees warfighters to focus on objectives rather than coordination. “People may not have enough time to [coordinate] because of other responsibilities,” Ortiz notes. “If you’re getting shot at, you can’t worry about what your schedule is going to be 10 minutes from now or an hour from now.” Using the software also reduces the human error that can occur when decisions are made during stressful situations.

The troops benefit in other ways as well. As they deal with the dynamics of a changing mission, they have limited time to make decisions. Instituting a major change five minutes before an event does not allow enough time to prepare. Coordinators can make alterations and alert the necessary parties quickly.

In addition, an individual Coordinator can adjust to its user. Because people have various work styles, a Coordinator can tailor itself to particular needs. 

Wagner believes the Coordinators project has multiple applications as coordination is a ubiquitous problem. “In this program we are focused on tactical field applications, but the need to coordinate action across multiple parties appears in the military at all levels,” he states. “There are also civilian applications for this technology.” Nonmilitary uses could include crisis response.

The program, still in development, begins the hands-on evaluation phase in 2007. During the testing, input from users will be used to perfect the product—for instance, how users will be alerted to schedule changes. Currently, a message appears on screen and the change is made to the schedule. In the future, an alert feature could be added, depending on user input.

“While this program is not developing new portable computing devices, the vision is that eventually soldiers in the field will interact with their Coordinators via natural interfaces such as head-up displays,” Wagner declares.

Developers also are working on a feature that takes military structure and protocol into account when making changes. “Eventually, military decision-making policies and procedures (doctrine) will be encoded for the Coordinators to use,” Wagner says. “This will enable them to reason about a given decision and to ensure that both the soldier and the Coordinator follow required notification protocols and make only decisions that are within their scope of authority.”


Web Resources
Defense Advanced Research Projects Agency:
University of Southern California Information Sciences Institute:
Honeywell International Incorporated:
SRI International:


Enjoyed this article? SUBSCRIBE NOW to keep the content flowing.