03/27 : Yet Another PoM/Prism Discussion¶

Participants:
Andrea, Ellon, Pierre, Quentin

Goal of the meeting: Discuss implementation details of PoM/Prism and try to take a decision about how to proceed.

Discussion¶

1. About accessing the time

• We agreed that we should have a core library managing the time. This library would be used by all DFNs and DFPCs that wanted to get the current time.
• This is in opposition to have PoM/Prism publishing timestamps on topics that would be used by other modules. This simplifies the access to time, and removing some complexity of the system to be designed
• If we will need time synchronized between multiple robots the time library would be updated with mechanisms allowing time synchronization. The interface with the modules would be the same.

2. About interfaces

• There was some discussion about how to design our code (in general, not only related to POM/Prism).
• We agreed that we should design our code on layers, from core ASN.1-free libraries, to ASN.1 dependent layers (probably the DFNs and DFPCs) and up to the modules/nodes that relate with the middleware, more or less how it is shown in the figure below.

• ADD FIGURE

• There was no agreement if bitstreams should be included or not on the ASN.1 layers. Quentin and Pierre are mostly in favour of the addition, while Andrea and Ellon were against it.
• We agreed that this was not a big problem if interfaces for both ASN1 C-structs and ASN.1 bitstreams were implemented.

3. About PoM/Prism design

• Recap of what we want for PoM:
  • PoM = Prism + Mighty Localizer
  • Prism
    • Maintain internal frames
    • Export frames (How? TBD)
    • Receives "Joint States"
  • Mighty Localizer
    • Receives robot poses at given times (from localization modules)
    • Responsible for fusion of poses
    • Exports the latest robot pose
    • Maintain/Manages past robot poses

• There were two main propositions, given that time access was simplified by assuming the existence of a time library.
• Proposition #1:

• ADD FIGURE

• Proposition #2:

• ADD FIGURE

4. About using genom-ros or only ROS

Open questions¶

• Where to put the interface between ASN.1 C-structs and ASN.1 bitstreams?