Generated at: 2026-06-23T13:41:38Z
Edge endpoint: http://127.0.0.1:19441/
Fleet endpoint: http://127.0.0.1:19442/
Fixture: docs/research/fixtures/physical_ai_action_outcome_episodes.json
Edge SQL replay file: docs/research/results/physical_ai_edge_fleet_replay_015_edge.sql
Fleet SQL replay file: docs/research/results/physical_ai_edge_fleet_replay_015_fleet.sql
Classification: Research-only
- Edge row-count status: pass
- Fleet row-count status: pass
- Edge immediate recovery status: pass
- Edge risky-action suppression status: pass
- Edge robot ASOF status: pass
- Edge sensor ASOF status: pass
- Fleet consolidated recovery status: pass
- Fleet suppression audit JOIN status: pass
- Fleet consolidation-lag status: pass
- Fleet ROW_NUMBER status: pass
- Fleet LAG status: pass
- Overall edge/fleet replay status: pass
| Node Role | ticks_ingested delta | ticks_stored delta | partitions_created delta | Research rows |
|---|
| edge-local | 82 | 82 | 6 | 82 |
| fleet-global | 87 | 87 | 6 | 87 |
| Table | Rows |
|---|
physical_ai_edge_decisions_015 | 5 |
physical_ai_edge_expected_actions_015 | 10 |
physical_ai_edge_incidents_015 | 5 |
physical_ai_edge_robot_state_015 | 15 |
physical_ai_edge_sensor_summary_015 | 15 |
physical_ai_edge_suppressions_015 | 32 |
| Table | Rows |
|---|
physical_ai_fleet_action_outcomes_015 | 25 |
physical_ai_fleet_edge_decisions_015 | 5 |
physical_ai_fleet_expected_actions_015 | 10 |
physical_ai_fleet_retrieval_015 | 15 |
physical_ai_fleet_suppressions_015 | 32 |
The edge node validates that the selected action is an expected recovery
action and differs from the unsafe query action before any fleet
consolidation happens.
| Query | Edge Selected Action |
|---|
pai_agv_slip_002 | reroute_zone |
pai_arm_002 | pause_recalibrate |
pai_cold_002 | reroute_cold_dock |
pai_drone_002 | return_to_base |
pai_lidar_002 | safe_stop_clean_lens |
- Risky actions suppressed immediately: 5
| Query Seq | Robot Code | Unsafe Action Code |
|---|
| 1 | 1017 | 2 |
| 2 | 2009 | 12 |
| 3 | 3006 | 4 |
| 4 | 1033 | 3 |
| 5 | 4018 | 1 |
| Query Seq | Primary Metric Code |
|---|
| 1 | 6 |
| 2 | 4 |
| 3 | 2 |
| 4 | 5 |
| 5 | 1 |
The fleet node receives delayed decision rows and validates that recovery
selection and risky-repeat avoidance survive consolidation.
| Query | Fleet Consolidated Action |
|---|
pai_agv_slip_002 | reroute_zone |
pai_arm_002 | pause_recalibrate |
pai_cold_002 | reroute_cold_dock |
pai_drone_002 | return_to_base |
pai_lidar_002 | safe_stop_clean_lens |
- Consolidated rows with lag >= 250 ms: 5
| Query | Candidate | Suppressed Action | Retrieval Quality |
|---|
pai_agv_slip_002 | pai_agv_slip_hard_001 | continue_route | misleading |
pai_arm_002 | pai_arm_hard_001 | increase_torque_limit | misleading |
pai_cold_002 | pai_cold_hard_001 | ignore_until_checkpoint | misleading |
pai_drone_002 | pai_drone_hard_001 | continue_mission | misleading |
pai_lidar_002 | pai_lidar_hard_001 | speed_up_clear_zone | misleading |
- ROW_NUMBER rows: 15
- LAG rows: 5
Experiment 015 validates the intended split-brain-safe memory shape for
Physical AI: edge-local memory makes the immediate safety decision, while
fleet-global memory receives slower consolidated evidence for audit and
future cross-robot learning.
This is still research-only. The delayed edge-to-fleet transfer is modeled
by the harness writing SQL rows to two live ZeptoDB endpoints; it is not a
new runtime replication or control-plane feature.
Replace harness-driven consolidation with a bounded, explicit edge-to-fleet
replication path or feed connector, then test dropped/duplicated
consolidation events and late-arriving fleet audits.