concept

AGV

Also known as: AGV, AGVs

Facts (25)

Sources

Leveraging Knowledge Graphs and LLM Reasoning to Identify ... arxiv.org arXiv Jul 23, 2025 25 facts

measurementThe proposed method identified that Forklift FL_00 had a package movement time from AGV to storage of 152 seconds, compared to a global average of 123 seconds.

claimThe ‘Step-wise Guide’ method succeeds in calculating AGV travel metrics because its structured decomposition allows for accurate identification of journey segments and correct aggregation of data from the knowledge graph.

codeThe Cypher query to calculate average forklift waiting time and utilization across all blocks during the discharge flow is: MATCH (a:AGV)-[]->(fl:FL)-[]->(st:S) RETURN fl.id, avg(wait_time), SUM(active_FL)/total_FL_span AS util

claimThe ‘Direct QA’ and ‘Direct QA + SR’ baselines failed to determine the correct average travel time for an AGV, likely due to misinterpreting the scope of the travel path or using incorrect aggregation methods over the knowledge graph.

referenceThe proposed framework utilizes a custom Knowledge Graph (KG) schema where resources such as suppliers, workers, AGVs, forklifts, and storage are represented as nodes, while the movement of packages between these resources is represented as edges. Operational data, including timestamps, is incorporated as features of these nodes and edges, with the KG constructed from output logs generated by a Discrete Event Simulation (DES) model.

codeThe Cypher query to calculate AGV utilization is: MATCH (w:WORKER)-[]->(a:AGV)-[]->... RETURN a.id, SUM(dur_proc) / dur_active AS util

claimThe self-reflection mechanism in the ‘Direct QA + SR’ baseline is insufficient to correct errors in calculation logic or path identification when determining AGV travel times.

measurementThe ‘Direct QA’ baseline calculated the average travel time for an Automated Guided Vehicle (AGV) to move a package from the dock to its assigned storage area as approximately 178 seconds.

claimThe baseline approach for analyzing warehouse slowdowns between 10:00 AM and 12:30 PM provided a general explanation based on average operational durations of workers (58s), AGVs (474s), and forklifts (118s), but failed to identify specific causes for the deviation.

measurementThe ‘Step-wise Guide’ method calculated the average travel time for an Automated Guided Vehicle (AGV) to move a package from the dock to its assigned storage area as 455 seconds.

codeThe Cypher query to calculate average package waiting times at the waiting point for each AGV during the discharge flow is: MATCH (w)-[wta:WTA]->(a:AGV) RETURN a.id, avg(dur(wta.worker_end, wta.agv_arrival)).min UNION ... global_avg

codeThe Cypher query to calculate the average time taken by each forklift to move a package from the AGV to the storage bay is: MATCH (a)-[]->(fl:FL)-[]->(st:S) RETURN fl.id, avg(FL_move) AS t1; ... global_avg AS t2 RETURN fl.id, t1, t2, t1-t2

codeThe Cypher query to calculate forklift utilization is: MATCH (a:AGV)-[]->(fl:FL)-[]->... RETURN fl.id, SUM(dur_proc) / dur_active AS util; ...avg(util)

claimThe researchers created 25 distinct operational questions to assess the proficiency of the system in retrieving factual information and performing analyses using simulation output, with a uniform distribution across entities like supplier interactions, worker activities, AGV and forklift utilization, and package flow.

measurementThe ‘Direct QA + SR’ baseline calculated the average travel time for an Automated Guided Vehicle (AGV) to move a package from the dock to its assigned storage area as approximately 178.14 seconds.

procedureThe procedure to determine AGV utilization during the CamelCargo unload process in the Discharge Flow involves calculating the total processing time divided by the total active time for each AGV handling CamelCargo packages, then comparing this result to the global average AGV utilization.

procedureThe LLM agent calculates waiting times for AGVs by subtracting the worker pick-up end time from the AGV arrival time, and for forklifts by subtracting the AGV journey end time from the forklift placement start time.

procedureThe simulation captures process and equipment-specific data, including equipment ID, arrival time, process initiation time, waiting time, and process completion time for suppliers, workers, AGVs, and forklifts.

accountDuring a warehouse discharge operation observed between 10:00 AM and 12:30 PM, a human expert noted that AGV operational times were longer for most packages between 10:30 AM and 11:11 AM, though it remained inconclusive whether the delay was caused by AGVs, specific workers, or forklifts.

measurementForklift FL_00 is identified as the primary bottleneck in the discharge flow, with an average waiting time of approximately 333 seconds and a package movement time from AGVs to storage of 152 seconds, which is 29 seconds above the global average.

procedureThe simulation captures package-specific data, including a unique package ID and timestamps for pickup from the supplier, waiting time at the transfer point, loading and departure time with an AGV, arrival time at the storage block, and final placement into storage.

measurementThe global average time for an AGV to transport a package from the waiting point to the pickup point is 8.7 seconds, with AGV_07 and AGV_09 performing faster than the average, and AGV_00 and AGV_08 performing slower than the average.

claimInefficiencies in 'AuroraFarms' operations were attributed to worker utilization rates as low as 2.6% and variable AGV (Automated Guided Vehicle) utilization rates as high as 86% during the observed timeframe.

measurementThe utilization rates for workers and AGVs (Automated Guided Vehicles) for AuroraFarms between 10:00 AM and 12:30 PM showed variation, with some worker utilization as low as 2.6% and some AGV utilization as high as 86%.

referenceThe knowledge graph schema for the warehouse simulation defines nodes for SUPPLIER (properties: supplier_id, arrival_time, discharge_start, discharge_end), WORKER (property: worker_id), AGV (property: agv_id), FL/Forklift (property: forklift_id), and STORAGE (property: block_id).