A ROS package containing custom ROS messages for communication with SAR games.
The GameCommand message is used to issue commands to SAR games. This message includes the following fields and a set of constants for these fields:
game: A constant listing which game should pay attention the command.
TODO: This field is still under experiment. The available games are:
command: Specifies the command for the receiving game. This command could
be any of the following (use cases described below):
level: Include when sending a GameCommand.START message. Specifies the
level the game should start at.
Upone receiving each command, the game should...
GameCommand.START: Start to play.
GameCommand.CONTINUE: Resume play. This command is expected to be send after
a GameCommand.PAUSE command.
GameCommand.PAUSE: Pause game play until a GameState.CONTINUE message is
received.
GameCommand.END: Exits gameplay gracefully. That is, continue to play until
the next reasonable exit point. Do not exit the game abruptly.
GameCommand.WAIT_FOR_RESPONSE: Wait for a player response before continuing.
If the game receives a player response, it can continue as normal; if it does
not, it should send a GameState.USER_TIMEOUT message to indicate that no
response was received.
GameCommand.SKIP_RESPONSE: Resume play, optionally taking additional action
to reflect the fact that an expected user response was not received.
The GameState messages are sent by games to indicate what's happening in the game. This message includes the following fields and a set of constants for these fields:
game: The same as for GameCommand, described above.
state: Set to one of the following constants (use cases listed below):
performance: Include when you send a GameState.END message. Should
contain the player's sucess rate(s) for the game's primary target skill (e.g.,
ordering/sequencing, perspective taking, emotional understanding) as a
decimal percentage (i.e., the number of successes or correct answers divided
by the number of opportunities for success). The perfomance field should be
a JSON string with names of performance metrics mapped to their associated
values.
The possible performance metrics that each game can report are as follows:
Rocket-Barrier Game:
child-explainer-location-accuracy : In the version of the game where
the child explains the rocket for the parent to build, the proportion of
pieces of the parent's constructed rocket that are correct in both piece
type and position.child-explainer-piece-choice-accuracy : In the version of the game
where the child explains the rocket for the parent to build, the
proportion of pieces on the rocket that are correct in piece type, but
not necessarily piece position.child-builder-location-accuracy : In the version of the game where
the parent explains the rocket for the child to build, the proportion of
pieces of the child's constructed rocket that are correct in both piece
type and position.child-builder-piece-choice-accuracy : In the version of the game
where the parent explains the rocket for the child to build, the
proportion of pieces on the child's constructed rocket that are correct
in piece type, but not necessarily piece position.Storytelling Game:
child-emotion-question-accuracy : The percentage of emotion questions
that the child answered correctly this session. These multiple-choice
questions ask the child to identify which emotions were felt by
characters in the stories.child-tom-question-accuracy : The percentage of theory of mind (ToM)
questions that the child answered correctly this session. These
multiple-choice questions ask the child to identify what emotion a
character in a story might feel next.child-order-question-accuracy : The percentage of order questions that
the child answered correctly this session. These multiple-choice
questions ask the child to identify in which scene of the story different
events happened, such as what happened first or when a particular
character felt sad.Galactic Traveler:
attempts_mean : The average number of attempts across the 5 activities.attempts_variance : The variance on the number of attempts.time_spent_mean : The average number of seconds the child spent on the
5 activities.time_spent_variance : The variance on the time spent on an activity.activity_with_most_attempts : The activity (1-5) that the child had the
most attempts.activity_child_spent_longest : The activity (1-5) that the child spent
the longest on.child_success_percentage : The percentage of activities the child
answered correctly without any failed attempts.Spaceship Tidyup:
attempts_mean : The average number of attempts across the 5 activities.attempts_variance : The variance on the number of attempts.time_spent_mean : The average number of seconds the child spent on the
5 activities.time_spent_variance : The variance on the time spent on an activity.activity_with_most_attempts : The activity (1-5) that the child had the
most attempts.activity_child_spent_longest : The activity (1-5) that the child spent
the longest on.child_success_percentage : The percentage of activities the child
answered correctly without any failed attempts.The following is an example of the performance field for the Rocket-Barrier game:
{
"child-explainer-location-accuracy" : 0.8,
"child-explainer-piece-choice-accuracy" : 0.6
}
The following is an example of the performance field for the Storytelling game.
Note that depending on the child's level in the Storytelling game, not all
types of questions will be asked. The performance metric for a question type
will only be reported if that type of question was asked that session. Thus,
sometimes only the child-emotion-question-accuracy will be reported,
sometimes both the child-emotion-question-accuracy and
child-order-question-accuracy will be reported, and sometimes all three will
be reported.
{
"child-emotion-question-accuracy" : 0.7,
"child-tom-question-accuracy" : 0.0,
"child-order-question-accuracy" : 0.5
}
For Galactic Traveler and Spaceship Tidyup, all performance metrics will be reported, as in the following example:
{
"attempts_mean" : 1.6,
"attempts_variance" : 0.64,
"time_spent_mean" : 5.125324999999999,
"time_spent_variance" : 7.4272715651532,
"activity_with_most_attempts" : 1,
"activity_child_spent_longest" : 1,
"child_success_percentage" : 0.6
}
The following empty json string should be sent in the performance field if the game ends and there is no performance information to report (rather than sending nothing in this field).
{ }
A game should publish GameState messages at the following times:
GameState.START: When a GameCommand.START message has been received and the
game is starting.
GameState.IN_PROGRESS: When a GameCommand.START, GameCommand.CONTINUE, or
GameCommand.SKIP\_RESPONSE message has been received, to indicate that the
game is continuing play now.
GameState.PAUSED: When a GameCommand.PAUSE message has been received and
the game is paused.
GameState.USER_TIMEOUT: If a response was expected from the user, but no
response was received within a reasonable amount of time. The game should PAUSE
after it sends this message to wait for instructions regarding whether it
should wait for a response again or skip the response and move on.
GameState.END: After wrapping up the game, when gameplay has ended.
GameState.READY: When a game is up and running and is ready to receive a
GameCommand.START message (when the game is ready to be started and played).