Skip to content

Latest commit

 

History

History
99 lines (93 loc) · 4.36 KB

TODO.md

File metadata and controls

99 lines (93 loc) · 4.36 KB
Raw

TODO List

##Action Plan##

  1. Write machining code using jscam
    • Get usage experience
    • Manual everything(feed rate etc.)
    • Reproduce existing gcode programs
  2. ...?

High Level

  • Longer term design goal (on branch postp)
    • Change from current design into independant tools
    • Have a tool that reads gcode and outputs structures objects
    • Simulator works on this structured objects - not tied to machine object.
    • Plain gcode could be analysed.
    • Implement gcode parser.
  • Simulation
    • Ensure feed rate passed to simulation is normalised.
    • Implement additional simulation steps.
      • Calculate machine force vectors
      • Calculate cutter engagements
      • Calculate machine torque required
    • Modify Machine interface to return simulation results
      • Current thought is to wrap simulation result API in a different machining result struct.
      • For Linear & Arc (Maybe rapid if useful information is calculated).
      • Split simulation data and methods between public and private headers
      • Simulation.h and Simulation_int.h
      • Requires split / move of Path.h and Path_int.h
  • Path expansion
    • Use path::expand_rotary in Machine when pure rotary motion is detected.
    • UVW mapped into cartesian space?
  • Check for stock intersection for Linear & Rapid movements
    • Rapids (Positioning)
      • Rotate tool along all angular axes
      • glide (minowski sum) rotated tool with box formed by linear axis movement
      • Perform intersection tests.
    • Linear / Arcs (Material removal)
      • Output
        • Chip load per tooth
        • Cutter engagement
        • Cutting time.
      • Performance
        • How to measure cutting performance?
        • Path for each flute is trochoidal.
        • Calculate the path that flute tip passes through material.
        • Gives simulated chip load per tooth.
        • add experiment for flute path
  • Higher order curves
  • Complete Tool class - apt_cutter branch
  • Complete Stock class
    • Needs material properties
  • Plunge motion
    • Will be implemented as primitve (allows optimisation and clarification of intent)
  • Ability to explode stock into individual objects when cut
    • How should user specify which part remains?
    • geom::explode not needing intermediate Polyhedron_3 objects (spliting based on Nef shells alone)
  • geom
    • polyhedron_t
      • Store and update Polyhedron_3
      • Return copy of Nef_Polyhedron_3
    • Meshing
      • Current meshing parameters are too computationally expensive.
    • Geometry
      • Validate primitive geometry
      • Correcting narrowing warnings.
    • Wrapper for CGAL/AABB_tree
      • Needed for motion planning (ray intersection tests)
    • Wrapper for CGAL/Mesh objects
      • Tetrahedral mesh desired to have better js api
  • Represent workholding, clamps, etc. (for intersection tests)
  • Restore position
    • Ensure move is safe (no intersection with tool or clamps)
    • First version rapid to clearance plane, move, then rapid back to previous z height.
  • Automatic/heuristic tool selection
  • Global Machine accuracy
    • Use when generating tool and stock models.
  • Math
    • Correct faulty equality ops for math::point_3 and math::vector_3.

Code

  • Ensure exceptions do not change state (to allow recovery)
  • Add Interface stability test.
  • Split tests (one per source file)
  • Add expected output to tests and verify.
  • User boost::operators to define operatiors on types.
  • Review boost::test for unit tests
  • Axis / Offset UDL type.
  • Tool::Model() -vs- Stock.Model

Notes

  • tool wear can be approximated as a cost for each mm linearly travelled through material for each tooth.
  • i.e. model troicoidal path that each tooth takes, measure length of curve from first contact with material to last (possibly use a gradient function to map cost - higher at contact and lower at exit (climb; opposite for conventional) then sum those costs as the tool wear cost.
  • Thought is that this will model tool wear reasonably well - tool wears at edge - repeated shallow cuts will cost more than fewer deeper cuts (relative to MRR).
  • It should be possible to measure chatter with a contact microphone.