Meeting

Attendees: @kartmann @weiner @pohl @reister @jacob @ujrwq_arlt @dreher (notes)

• @dreher: Discussion new Axii modules for the new repositories (names TBD)
• @kartmann, @dreher: Presentation of first draft of the class diagramm
• Discussion from discussion of class diagram:
  • Observation: Differentiation between Sensor and ControlTarget (not Actuator!) is good and should be used, as this is more honest
  • An Actuator is still required, as these "bundle" ControlTargets which are mutually exclusive, since we do not have a "slave" concept in the framework.
  • Actuators can be virtual (e.g., as control target for virtual controller such as the platform) or a real participant on the bus (e.g., an Elmo)
• Scheduler: Arbiter concept required to keep track of which control targets are already used.
• Calibration: Slaves will do it and pass correctly initialized objects (from hw config) to the bus.
• Discussion of "Realtime Robot Model" as interface
• Disucssion about dependencies (Simox and "QUantities" library)
• @weiner, @pohl: Discussion sequence diagram

Tasks remain more or less as previous week.

Identified invariants for rtRun:

• In each cycle, the control targets are "nulled", and need to be explicitly set (make sure that controllers are still alive and acting)
• Scheduler: Keep track of requested actuators and make sure that each actuator is assigned to a controller