Come up with consistent notation

[spmaniato]

@pschillinger and @dcconner After skimming through the paper, could you offer suggestions for how to make the nomenclature and notation more consistent please?

The way I see it, there are a bunch of things going on:

• robot's capabilities and the capabilities provided by our software
  • I've been saying that system = ALTAS + Team ViGIR software
• software components (e.g. MoveIt)
• atomic capability (e.g. the ability to open/close fingers)
• (parametrized) primitive (e.g. close robotiq fingers 90%, change control mode to STAND)
• outcomes of primitives and outcomes of propositions
• "state machine" synthesis vs "behavior" synthesis
  • In the video/screenshots, it looks like I'm synthesizing a state machine
  • A "high-level behavior" and a "FlexBE behavior" are different concepts
• abstract automaton / finite state machine vs FlexBE / SMACH state machine (Python)
• state of some state machine vs FlexBE state (e.g. one of the vigir_flexbe_states)
  • From a synthesis point-of-view, the name "state" only makes sense in the context of a SM
  • But we want a way to refer to what we've been calling states. Maybe "state primitives"?
• userdata (which I want to avoid discussing in length, or at all, in this paper)

[dcconner]

I"m swamped with course/lecture prep, plus fighting the final report through next week. I'm hopefully it will calm down at end of week to let me look over this paper and other

[dcconner]

Not to mention getting up to speed on Mac! Got SourceTree to let me access GItHub, but not sure what to do for latex. It seems that my computer is set to block all but app store by default, and I didn't see any latex compilers on app store.

[spmaniato]

OK, no worries. Btw, last week I discovered that Github can visualize PDFs in the browser. See for example main.pdf Very convenient for taking quick looks at the compiled paper.

[pschillinger]

Could you point us to some examples where the nomenclature is unclear/ambiguous? I think some context will help for coming up with working consistent expressions.

You can refer to FlexBE's states (in contrast to the abstract term "state" in the SM-context) as either state implementations (Python class) or state instantiations (Python object used in an SM realization). At least that's how I used to do it.

I would say that the term "primitive" is quite broad and can be used most of the time when writing about something the robot is able to do without Synthesis/FlexBE involved.

What exactly do you mean by "outcome of propositions"? A proposition can either be true or false, right? So do you refer to its truth value / boolean value? Or is it in the context of an boolean implication?

[spmaniato]

What exactly do you mean by "outcome of propositions"? A proposition can either be true or false, right? So do you refer to its truth value / boolean value? Or is it in the context of an boolean implication?

Sorry, I meant "outcome propositions" (without "of"). There are activation propositions (True when robot activates a primitive) and outcome propositions (True when the corresponding outcome, e.g., "planned", is returned).

Could you point us to some examples where the nomenclature is unclear/ambiguous? I think some context will help for coming up with working consistent expressions.

One (not so significant) example is the notation for the "outcomes of x". I use Out(x) while in your thesis it's x_Oc or something like that.

Most conflicts haven't arisen yet because I had been focusing on the LTL section. Once we start connecting the pieces they'll come up. I'll get back to you with concrete examples.

[spmaniato]

As of 18f76a71212e77b1aa23ccfddbb60a3281662cd7:

• System = Atlas + Team ViGIR software
• Behavior is an abstract term, not the top-level FlexBE state machine
• Capability = low-level functionality (left kind of ill-defined on purpose)
• State implementation = the Python classes that developers write (as per Philipp's thesis)
• Parametrized state = a state implementation with its parameters set
• No mention of userdata

[spmaniato]

I made some more changes yesterday. Overall I think I've converged to a pretty accurate and consistent notation. The only exceptions are the term "behavior" and this magical thing we call "primitive capabilities", C, which I even use in mathematical formulas without ever really defining ...