Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Knowledge-Graph Translation Layer for Mission-Aware Multi-Agent Path Planning in Spatiotemporal Dynamics

Edward Holmberg, Elias Ioup, Mahdi Abdelguerfi

TL;DR

The paper introduces a knowledge graph based translation layer that decouples high level mission semantics from a domain agnostic path planner by producing mission aware worldviews and physics aware seam rules through a two plane architecture. The data plane compiles semantic inputs into per agent heatmaps while the control plane provides dynamic stitching costs, enabling an MPC like incremental replanning loop. A Gulf of Mexico AUV case study demonstrates that declarative changes to KG policies yield distinct, high performing plans, highlighting the framework's explainability and adaptability. The approach advances autonomous system design by treating the KG as a stateful orchestrator for end to end planning, monitoring, and replanning in dynamic environments.

Abstract

The coordination of autonomous agents in dynamic environments is hampered by the semantic gap between high-level mission objectives and low-level planner inputs. To address this, we introduce a framework centered on a Knowledge Graph (KG) that functions as an intelligent translation layer. The KG's two-plane architecture compiles declarative facts into per-agent, mission-aware ``worldviews" and physics-aware traversal rules, decoupling mission semantics from a domain-agnostic planner. This allows complex, coordinated paths to be modified simply by changing facts in the KG. A case study involving Autonomous Underwater Vehicles (AUVs) in the Gulf of Mexico visually demonstrates the end-to-end process and quantitatively proves that different declarative policies produce distinct, high-performing outcomes. This work establishes the KG not merely as a data repository, but as a powerful, stateful orchestrator for creating adaptive and explainable autonomous systems.

A Knowledge-Graph Translation Layer for Mission-Aware Multi-Agent Path Planning in Spatiotemporal Dynamics

TL;DR

The paper introduces a knowledge graph based translation layer that decouples high level mission semantics from a domain agnostic path planner by producing mission aware worldviews and physics aware seam rules through a two plane architecture. The data plane compiles semantic inputs into per agent heatmaps while the control plane provides dynamic stitching costs, enabling an MPC like incremental replanning loop. A Gulf of Mexico AUV case study demonstrates that declarative changes to KG policies yield distinct, high performing plans, highlighting the framework's explainability and adaptability. The approach advances autonomous system design by treating the KG as a stateful orchestrator for end to end planning, monitoring, and replanning in dynamic environments.

Abstract

The coordination of autonomous agents in dynamic environments is hampered by the semantic gap between high-level mission objectives and low-level planner inputs. To address this, we introduce a framework centered on a Knowledge Graph (KG) that functions as an intelligent translation layer. The KG's two-plane architecture compiles declarative facts into per-agent, mission-aware ``worldviews" and physics-aware traversal rules, decoupling mission semantics from a domain-agnostic planner. This allows complex, coordinated paths to be modified simply by changing facts in the KG. A case study involving Autonomous Underwater Vehicles (AUVs) in the Gulf of Mexico visually demonstrates the end-to-end process and quantitatively proves that different declarative policies produce distinct, high-performing outcomes. This work establishes the KG not merely as a data repository, but as a powerful, stateful orchestrator for creating adaptive and explainable autonomous systems.
Paper Structure (60 sections, 9 figures, 7 tables)

This paper contains 60 sections, 9 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Spatiotemporal Points of Interest (POIs)
  4. Dynamic Environmental Conditions
  5. Multi-Agent Coordination
  6. Knowledge Graphs for Representing Complex Systems
  7. Related Work
  8. Path Planning in Dynamic Environments
  9. Multi-Agent Coordination and Task Allocation
  10. Knowledge Graphs in Robotics and Planning
  11. Methodology
  12. The Knowledge Graph Model
  13. The Two-Plane Translation Architecture
  14. The Data Plane ($\Phi_2$): Compiling Mission Tensors
  15. The Control Plane ($\Phi_1$): Governing Coordination and Selection
  16. ...and 45 more sections

Figures (9)

  • Figure 1: Architectural overview of the Knowledge Graph as a central orchestrator. The Data Plane ($\Phi_2$) compiles semantic inputs into a stack of numerical Mission Tensors, which are fused into an Aggregate Tensor for the Agnostic Path Planner. The planner generates Pathlet Candidates. The Control Plane ($\Phi_1$) then orchestrates the selection and coordination from these candidates to produce the final Coordinated Mission Plan. Feedback from rejected paths can inform subsequent compilations, creating an adaptive planning loop.
  • Figure 2: High-level class hierarchy (ASCII tree).
  • Figure 3: Per-agent work tensors at two poses (green=easy, red=costly).
  • Figure 4: Aggregated mission tensor (FUSED) from KG compilation (Data Plane $\Phi_2$).
  • Figure 5: Agent “worldview”: value (colors) with environmental work vectors.
  • ...and 4 more figures