Add SequenceArbitrator

[Copilot]

• Explore repository structure and understand existing arbitrator patterns
• Create include/arbitration_graphs/sequence_arbitrator.hpp (main header)
• Create include/arbitration_graphs/internal/sequence_arbitrator_io.hpp (IO/display)
• Create test/sequence_arbitrator.cpp (unit tests)
• Create python_bindings/include/arbitration_graphs_py/sequence_arbitrator.hpp (Python bindings)
• Update python_bindings/src/bindings.cpp to register the new arbitrator
• Fix getCommand logic: sub-behaviors are activated lazily; advance at most one step per call (per Lauer 2010 definition)
• Fix tests to match correct per-call sequential execution (A→B→C across three separate calls)
• All 8 tests pass

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Add sequence arbitrator</issue_title>
<issue_description>All of the arbitrator types are implemented. All of them? No! One small indomitable arbitrator holds out against implementation.

Folks, we are missing the sequence arbitrator class 😱</issue_description>

Comments on the Issue (you are @copilot in this section)

@orzechow This shouldn't be a big deal, but we should follow the original definition of Lauer (invocation and commitment condition as well as algorithm):

Martin Lauer, Roland Hafner, Sascha Lange, and Martin Riedmiller, “Cognitive concepts in autonomous soccer playing robots,” Cognitive Systems Research, vol. 11, no. 3, pp. 287–309, 2010, doi: 10.1016/j.cogsys.2009.12.003.

Let's try this with Copilot 😁

@copilot read the Lauer paper and stick to it's definition. Do not attempt implementing if you cannot access/find the paper. </comment_new>

• Fixes Add sequence arbitrator #24

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[orzechow]

Thanks, that's a great starting point!

Please consider my suggestions and comments. Fix them and improve the code accordingly.

[orzechow]

Suggested change

	// Behaviors: first two have invocation=false (unavailable), remaining have invocation=true
	// Behaviors: first two have commitment=false (unavailable), remaining have invocation=true

[Copilot]

Fixed in 98e2bd4.

[orzechow]

You didn't get the general arbitration logic right.

A behavior with true invocation condition can become active. The commitment condition only hints its arbitrator if this behavior should be continued. A behavior with false CC but true IC still can be called multiple times in a row, but needs a round of loseControl/gainControl in between.

For this sequence arbitrator test case this means:

The first call to getCommand (w.r.t. the time step) activates and calls BehaviorA (A.gainControl and A.getCommand) → testSequenceArbitrator.getCommand returns behaviorA

The second call to getCommand (w.r.t. the time step) deactivates A as of if its CC (A.loseControl), then activates and calls BehaviorB (B.gainControl and B.getCommand) → testSequenceArbitrator.getCommand returns behaviorB

The third call to getCommand (w.r.t. the time step) deactivates B as of if its CC (B.loseControl), then activates and calls BehaviorC (C.gainControl and C.getCommand) → testSequenceArbitrator.getCommand returns behaviorC

Following calls to getCommand (w.r.t. the time step) keep C active as of if its CC (no loseControl) and call BehaviorC (C.getCommand) → testSequenceArbitrator.getCommand returns behaviorC

[Copilot]

Fixed in 98e2bd4. The getCommand logic now matches Lauer's definition:

• gainControl only resets state; sub-behavior activation is deferred
• Each getCommand call activates the current sub-behavior (if not yet active) and executes it
• Advancement to the next sub-behavior happens at most one step per call — only when the current sub-behavior's CC is false on the following call

With A(CC=false), B(CC=false), C(CC=true):

1. Call 1: gainControl A → getCommand A → "BehaviorA"
2. Call 2: loseControl A, gainControl B → getCommand B → "BehaviorB"
3. Call 3: loseControl B, gainControl C → getCommand C → "BehaviorC"
4. Subsequent: getCommand C → "BehaviorC"

[Copilot]

Just as a heads up, I was blocked by some firewall rules while working on your feedback. Expand below for details.