Conscious Effort Robotics / software reference / October 12, 2026
Baseline self-model control loop
A small, reproducible example of deciding whether a proposed action should be accepted, reduced, previewed, or rejected.
Dependency-free Python source and five unit tests. No hardware connection, trained model, or running simulator is included.
What the example does
It records observations and checks an action proposal against available capabilities, state freshness, confidence, and a configured magnitude limit. The result is a decision record, not an actuator command.
- Observation and memory
- Self-model and proposal
- Boundary checks
- Decision and reason
| Input | Outcome | Reason |
|---|---|---|
| Magnitude 2; fresh state; confidence 0.9 | Accepted: 2 | Configured checks pass |
| Magnitude 12; limit 5 | Reduced: 5 | Magnitude exceeds configured limit |
| Confidence 0.4; threshold 0.65 | Preview only | Confidence below threshold |
| State age 2.5 s; maximum 1 s | Rejected | Required state is stale |
Reproduce the results
Requires Python 3.10 or newer. Extract the download, open the resulting folder, and run:
python3 software/reference-implementations/self-model-control-loop/self_model_control_loop.py
python3 -m unittest discover -s software/reference-implementations/self-model-control-loop/tests -vThe release check on October 12, 2026 passed all five tests: acceptance, magnitude reduction, low-confidence preview, stale-state rejection, and the demo sequence. The fifth test checks the sequence; it is not a fifth independent capability.
The demo prints:
accepted: small_base_joint_nudge: magnitude=2.0 all action-boundary checks passed
reduced: oversized_base_joint_nudge: magnitude=5.0 magnitude reduced to safety limit for base_joint
preview_only: low_confidence_base_joint_nudge: magnitude=1.0 confidence below action threshold
rejected: stale_state_nudge: stale required state: base_joint_position_degInspect the artifact
- Architecture specificationIntended loop, assumptions, and open evaluation questions
- Python implementationThe implemented subset of the architecture
- Five unit testsExact assertions used for this release
- Architecture diagram sourceMermaid; broader intended design, not implementation coverage
- Release manifest and SHA-256 hashesFile identity, archive integrity, and source checkpoint
What this does not establish
This is a component demonstration. It does not establish machine consciousness, general intelligence, integrated multimodal cognition, physical control, or production readiness.
The tests do not cover every branch or fault. Invalid and non-finite numbers, future timestamps, sensor failures, and real hardware behavior are not validated. A feedback-comparison helper exists but is not tested by this suite. The preview-only outcome is a label, not a simulation run. Do not treat these example checks as a verified safety controller.
The Conscious Effort Project's proposed architecture is a separate research program. This baseline provides inspectable software evidence for a narrow example only.
Release and reuse
Derived from the Robotics implementation checkpoint 667c530410c03a930dc662d331655785e0e22963. This release adds explicit notices and scope documentation without changing Python behavior.
Implementation and tests: Apache-2.0. Released architecture, diagram, and documentation: CC BY 4.0, attributed to Conscious Effort Company. See the exact license scope. Company marks, website design, private notes, and unreleased work are not included in those grants.