Multi-Agent · Pipeline

From Planner Logs to Explanations

MAPF Planneragents · paths · conflicts
→
JSON Logcanonical trace
→
generate_kg.pyPython · RDFLib
→
maPO Knowledge GraphTurtle / OWL / JSON-LD
→
SPARQL + NLexplanation templates
maPO system architecture
System architecture — planner-specific logs are mapped into a planner-agnostic, SPARQL-queryable knowledge graph.
Planner-specific inputRaw MAPF traces & optional args — not a custom explanation API.
Ontology mappinggenerate_kg.py asserts RDF triples automatically.
Planner-agnostic outputOne graph drives queries, replay & explanations.
python generate_kg.py --log_file <json> --output <ttl>