Skip to content

Processor motion composite state adaptation

For the moment, each processor motion update its last frame by a call to the global problem getState which may be an issue for several reasons:

  • if several processor motion propagate the same stateblocks, the last frame of a lower priority processor might be updated by another processor
  • there is no guarantee that a propagated value exist for each state block of the problem structure at the time of the last frame state update

Benefiting from the new composite state API, it would be better to update only the last frame state blocks that are propagated by the processor motion. Then, when a processor motion receives a KF callback, it has to update the KF state blocks with its own corresponding propagated state blocks.

Example:
Proc IMU: propagates POV
Proc FTPreint: propagates CDLO
When proc IMU creates a KF, its state is initialized at the latest timestamp for POV sbs but at the origin timestamp for CDL. When proc FTPreint receives this KF, it has to "fill" the CDL sbs with its propagated state at the pack timestamp.