Ra is the main program used for development of our AI scripts.
Ra can either use the builtin simulator or use the data provided by SSL-Vision. Theres also a small builtin refbox which can be used in place of a network referee.
The GUI is made up of various widgets which can (for the most part) be freely moved inside the window. To open the menu used to hide or show widgets, make a middle mouse click in the toolbar on the top.
Located in the center and can't be disabled. The field can be moved around by holding down the right mouse button and moving the cursor. Zooming is possible by scrolling. On MacOS, theres support for the usual gestures. While the simulator is active, the ball can be moved around with the left mouse button and teleported using the Control key and the left mouse button. Robots can be moved around by left clicking them and dragging the mouse. Use the context menu to change various settings:
- field orientation
- whether visualizations should be drawn
- give ball placement commands
Located on the left of the field widget and can't be disabled. Can show debug information provided by the loaded AI modules. Its possible to filter the shown output using the search bar on the top. Hover over it for further information.
Located on the right of the field widget by default. Allows drawing various
visualizations on the field widget. The AI script base library contains a
module vis which has a high level interface that can be used to draw common
shapes. Double clicking a robot in the field widget toggles all visualizations
whose name ends in ": <robot id>".
Located below the field widget and can't be disabled. Will show text output by the loaded AI modules alongside a timestamp. This is also the place where autoref decisions are displayed.
This widget allows loading an AI script for each team, for the internal autoref and to assign robots to a team. Ra will only try to send radio commands generated by an AI script if the robot is set to be part of the corresponding team. Its also possible to select a robot for manual control.
Allows sending referee commands which is useful during simulator play and test games. In order to allow an autoref script to be loaded in the Robots widget, tick the "Internal Autoref" checkbox.
Found on the top of the window and provides some central configuration switches. The toolbar is filled by the "Commands", the "Testing" and the "Configuration" widget. The following order only applies if all of them are enabled (which they are by default).
The first button shows a radio antenna which enables the transceiver for robot communication. In the context of this release it is only important while the simulator is active. The capacitor next to it enables charging in order to allow robots to kick the ball.
The fourth button, showing a computer display, is used to to switch between the internal simulator and SSl-Vision.
The whistle button is used to disable and enable the "Internal RefBox" widget.
Pressing the gamepad button allows robots to be controlled manually. They also have to be selected for manual control in the Robots widget.
The vision port can be configured in the settings dialog.
Ra provides various tools used to analyze games.
Ra can be either in normal mode or in logplayer/Horus mode. In Horus mode,
the field widget will transform to show a progress bar and will allow to load
logfiles. There are multiple workspaces available. To switch between them type
Alt-1, Alt-2, Alt-3... Odd numbered workspaces are in normal mode, even
numbered in Horus mode. These workspaces have their own widget configuration.
While in Horus mode, the search bar allows seeking to an arbitrary position in
the logfile. The playback speed can be changed by changing the 100% scroll box.
The playback can be controlled using shortcuts. Look at the menu items under
Logging and Instant Replay for a complete list. Playback can be
paused/continue by pressing Space.
To start recording a game press the red button on the toolbar. Logs will be
saved in the Ra process working directory by default, but another location can
be set using Select logfile default locations item of the logging menu.
Ra automatically records the last 20s played. To save this as a log, use the 20s backlog button on the toolbar.
While in Horus mode, a new button 20s (log) backloglog button will appear. While backlog will save the last real 20s of playtime, backloglog will save the last 20s of a replay (see next section).
Ra also supports Instant Replay. Pressing Ctrl-A will switch to a Horus
workspace with a log of the last 20s played. If the simulator is not active,
this is only possible during halt. To exit Instant Replay, either type Return
or switch to a workspace in normal mode.
Its possible to merge together multiple logfiles into one and remove unwanted
sections like halt. To do so, select the Log Cutter item in the File menu.
Ra allows opening SSL Vision logs. They will be converted to our own log format on the fly.
Ra allows re-running an AI script on a log. This is called Replay (not to be confused with Instant Replay). To do so load a logfile and load an AI script in the replay widget. The replay scripts output will be tagged, e.g. when you replay a yellow strategy, the output in the debug tree will be prefixed by "Yellow Replay".
Ra has a plotter that can either plot values provided by an AI script or information about robots and the ball. To access it, click the button with the red and blue line in the toolbar. You can adjust the scale of inputs and outputs using the input fields on the top and stop displaying new data using the "Freeze" button.
You can calculate slopes by clicking in the output window and dragging the mouse.
To perform a ball speed measurement set Y Max to 10 and Y Min to 0 and
select Ball/v_global (only available if the ball is/was visible). When the
ball is shot check whether its speed is higher than about 8,5 m/s.