|
Remote AgentHow does a remote agent work? |
Here is an authentic example which illustrates how the Planner/Scheduler works. During its mission, DS1 performs a task called autonomous optical navigation (OpNav) to determine its position in space. During this process, DS1's Miniature Camera and Imaging System (MICAS) takes several photographs of specific asteroids. The planning and scheduling for this event is more complex than it may seem. Let's see how the PS does it!
Obtaining the Mission Goals. In order to begin, the PS obtains the main goals for a particular part of the mission. These goals can be preprogrammed before liftoff or can be uploaded from ground control during the mission. These goals are divided into separate categories. One of these categories is "waypoints" Waypoints tell the PS the start time and end time for a plan. This period of time is called a "schedule horizon." Another category is OpNav. The goals show when during the mission OpNav activities should take place. During DS1's mission, OpNav will take place once every week.
Timelines: Each category of goals is laid out on a separate timeline. If we are planning on one OpNav activity lasting 4 hours during a 66 hour period, that activity would be put on the 66 hour timeline. Instead of having an exact time at which the event must happen, a window of time longer than 4 hours will be specified for the OpNav event. The event can happen any time during that window. Many goals have large windows of time like this in which the goal could be accomplished, rather than an exact time it must be accomplished.
Planning Experts. When you create your plan to go to the store, there are certain parts of your plan that you leave up to other "experts." For example, many things must take place for your car to run. Fuel must be released and combusted, releasing energy to turn your tires to make your car go. You cannot be expected to know these details; you rely on your car to figure this out. PS also relies on experts to figure out some details of the plan; these experts are programs called "planning experts." In our example, PS would gather information from the autonomous navigation software program (AutoNav) and the attitude control system (ACS). AutoNav would provide information on where the asteroids are that will be photographed and when the photographs should take place. The ACS will tell PS how much and in what direction to turn in order to face MICAS toward the target asteroids. As you can see, many parts of DS1 are constantly communicating with one another. Each one has its own area of expertise.
Subgoaling. It is now the job of the planner to break down the goals into smaller tasks, or subgoals that must take place in order for the goals to be accomplished. This process is called subgoaling. For example, the OpNav activity goal from the mission manager can be subgoaled into a series of turns and MICAS taking photographs. MICAS taking a photograph must then be further subgoaled. While subgoaling, PS takes into account some constraints or rules that govern when activities must take place and plans accordingly. For example, MICAS must be on before it can take a picture. Then PS generates timelines that describe things things that must happen with the various spacecraft systems to accomplish the goals.
Each of the actions or states in the plan above is called a token. The timelines are set up so that no two tokens overlap. Since "MICAS ready" and "MICAS take picture" happen at the same time, they must be on two separate timelines.
Constraints. . As stated above, many constraints must be taken into account when a plan like the one above is generated by PS.
Once the plan is generated, it is sent to EXEC to be carried out.
What exactly can Remote Agent do?
What is AutoNav?
What is an asteroid, a planetoid, a meteoroid, a micrometeoroid?
How does DS1 take pictures?
How does DS1's electrical system work?
What is attitude control?
