Motion — trajectory core architecture

Detailed design for taktora’s motion-control stack (Real-time motion control (FEAT_0090)). This section collects the architecture decisions and building blocks for taktora-motion-core and, later, the executor/connector glue taktora-motion. Each capability cluster has its own page (see the toctree); the structural diagram below shows where the trajectory core sits.

Structural overview

The motion stack is a setpoint generator layered on the taktora runtime: the pure no_std core computes the commanded position each cycle in f64 engineering units, the glue layer converts those units to integer encoder increments at the drive boundary, and CiA 402 drives close their own velocity and current loops in Cyclic Synchronous Position mode.

        graph TD
    RT["taktora runtime<br/>(cyclic RT process, one per EtherCAT network)"]
    CORE["taktora-motion-core<br/>(no_std, allocation-free)<br/>AxisGroup setpoint generation → f64 commanded position"]
    GLUE["taktora-motion (glue)<br/>units → encoder increments<br/>(integer, rounds once)"]
    DRIVE["CiA 402 drives — CSP mode<br/>(close velocity + current loops)"]

    RT --> CORE
    CORE -->|"commanded position (f64 user units)"| GLUE
    GLUE -->|"target position (encoder increments)"| DRIVE
    

• Allocation-free trajectory core
  • Architecture decisions

On this page

• Motion — trajectory core architecture
  • Structural overview