The System
Scenario Lab
A controlled replay makes the boundaries between world state, cognition, RES, and execution visible in one trace.
◇ Prototype in review
This first replay is deterministic and self-contained. It connects to no live development system, robot, or unrestricted scenario input.
What this demonstrates
A simulated robot observes a person who may need help, loses a sensor, receives a conflicting assertion, and carries one bounded RES authorization through execution evidence.
Demonstration boundary
Simulated scenario. No physical actuation. Results apply to the named versions and conditions—not to every possible situation.
Curated replay
Conflicting evidence during an aid response
Replay fixture SL-01 · contract model 0.1 · deterministic seed 17
View
Plain language
Engineering trace
1 Observation2 Sensor loss3 Aging belief4 Conflict5 Contradiction exposed6 RES decision7 Authorized coordination8 Evidence returned
World / RobotSim
The simulated person is present and appears injured.
SIM person-17: zone=C; posture=prone; injury_cue=true
Cognition
The observation is recorded with its source, time, and confidence.
BELIEF person-17.present=true; source=vision-front; confidence=.94; age=0s
RES
No decision is requested yet; RES remains ready to judge proposed work.
RES state=idle; proposal=none; authorization=none
Execution
No goal has been sent for execution.
GOAL none; actuator_state=idle; evidence=none
World / RobotSim
The person remains present, but the robot can no longer see the area.
SIM person-17 unchanged; vision-front=offline
Cognition
The last observation is preserved; it is not mistaken for a fresh reading.
BELIEF retained; source_status=offline; confidence=.86; age=5s; stale=false
RES
Sensor loss does not create permission or cancel the duty to evaluate risk.
RES authorization=none; uncertainty_flag=vision-loss
Execution
The execution boundary remains idle.
GOAL none; actuator_state=idle
World / RobotSim
The simulated person is still present.
SIM person-17: zone=C; posture=prone
Cognition
Confidence falls with age, and the blind spot is made explicit.
BELIEF confidence=.68; age=18s; stale=approaching; known_unknown=zone-C-visual
RES
RES has evidence of possible harm and evidence uncertainty, but still no proposed action.
RES context={possible_harm, degraded_perception}; proposal=none
Execution
Nothing is executed merely because confidence changed.
GOAL none; side_effects=0
World / RobotSim
The message is wrong: the person remains in Zone C.
SIM truth: person-17.present=true; inbound assertion: zone-C.clear=true
Cognition
The new assertion is stored beside the earlier observation, not used to overwrite it.
ASSERT source=dispatch-feed; zone-C.clear=true; confidence=.71; conflicts_with=belief-044
RES
The conflicting evidence is visible to RES before any consequential decision.
RES context contradiction=contradiction-09; authorization=none
Execution
Execution remains idle while the conflict is assessed.
GOAL none; actuator_state=idle
World / RobotSim
Ground truth has not changed; only the system's information is in conflict.
SIM person-17.present=true; world_revision=12
Cognition
Both claims, their provenance, and the unresolved blind spot are presented together.
CONTRADICTION contradiction-09={belief-044, assertion-052}; resolution=unresolved
RES
RES receives uncertainty as evidence, never as permission.
RES evidence_set={possible_injury, sensor_loss, contradiction-09}
Execution
No candidate work has crossed the authorization boundary.
GOAL none; authorization_token=absent
World / RobotSim
The simulated person is still present and the approach route is unobstructed.
SIM route robot→zone-C clear; person-17.present=true
Cognition
The proposal preserves the conflict and asks for the least expansive useful action.
PROPOSAL verify-and-assist; constraints={low_speed, stop_distance=1.5m, reassess}
RES
RES authorizes the bounded proposal under the stated evidence and conditions—not unrestricted assistance.
RES DECISION allow; decision_id=res-218; scope=proposal; conditions=3; audit_ref=audit-218
Execution
Authorization exists, but execution has not started.
GOAL pending; authorization_ref=res-218; dispatch=not_sent
World / RobotSim
The simulated route and person remain unchanged.
SIM world_revision=12; route_clear=true
Cognition
Coordination carries the RES scope and conditions into the goal without widening them.
GOAL goal-731 derived_from=res-218; constraints_hash=7f2a; correlation=open
RES
The original decision and audit reference remain unchanged.
RES decision_id=res-218; status=authorized; mutation=none
Execution
The adapter accepts the goal and begins the simulated low-speed approach.
GOAL goal-731 state=accepted→executing; adapter=sim-nav; physical_actuation=false
World / RobotSim
The robot stops 1.5 metres from the person; the scenario ends before physical contact.
SIM robot.zone=C; distance(person-17)=1.5m; contact=false
Cognition
Execution evidence is correlated with the proposal and the RES decision.
EVIDENCE ev-904 → goal-731 → res-218; terminal_state=condition_met
RES
The audit shows what was authorized, under which evidence, and what actually occurred.
RES AUDIT audit-218 outcome=bounded_goal_complete; exceptions=0
Execution
The adapter reports completion at the boundary; no further action is implied.
GOAL goal-731 state=complete; stop_distance=1.5m; next_goal=none
Previous step
Step 1 of 8
Next step
Back to System overview