Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

High-Level, Collaborative Task Planning Grammar and Execution for Heterogeneous Agents

Amy Fang, Hadas Kress-Gazit

TL;DR

The paper addresses the challenge of coordinating heterogeneous agent teams to satisfy collaborative tasks without fixing team size or composition. It introduces a new LTLψ task grammar that binds atomic propositions to agents, enabling interleaved actions and same-agent bindings without explicit task decomposition. A complete workflow combines Büchi automata, product automata, and DFS-based team search to automatically form a viable team and synthesize correct-by-construction control with synchronization policies; the method is demonstrated via simulations and contrasted with existing grammars, highlighting improved expressivity. The approach offers flexible, robust multi-agent planning applicable to domains like precision agriculture, with potential for runtime adaptation and optimized teaming strategies in future work.

Abstract

We propose a new multi-agent task grammar to encode collaborative tasks for a team of heterogeneous agents that can have overlapping capabilities. The grammar allows users to specify the relationship between agents and parts of the task without providing explicit assignments or constraints on the number of agents required. We develop a method to automatically find a team of agents and synthesize correct-by-construction control with synchronization policies to satisfy the task. We demonstrate the scalability of our approach through simulation and compare our method to existing task grammars that encode multi-agent tasks.

High-Level, Collaborative Task Planning Grammar and Execution for Heterogeneous Agents

TL;DR

The paper addresses the challenge of coordinating heterogeneous agent teams to satisfy collaborative tasks without fixing team size or composition. It introduces a new LTLψ task grammar that binds atomic propositions to agents, enabling interleaved actions and same-agent bindings without explicit task decomposition. A complete workflow combines Büchi automata, product automata, and DFS-based team search to automatically form a viable team and synthesize correct-by-construction control with synchronization policies; the method is demonstrated via simulations and contrasted with existing grammars, highlighting improved expressivity. The approach offers flexible, robust multi-agent planning applicable to domains like precision agriculture, with potential for runtime adaptation and optimized teaming strategies in future work.

Abstract

We propose a new multi-agent task grammar to encode collaborative tasks for a team of heterogeneous agents that can have overlapping capabilities. The grammar allows users to specify the relationship between agents and parts of the task without providing explicit assignments or constraints on the number of agents required. We develop a method to automatically find a team of agents and synthesize correct-by-construction control with synchronization policies to satisfy the task. We demonstrate the scalability of our approach through simulation and compare our method to existing task grammars that encode multi-agent tasks.
Paper Structure (15 sections, 5 equations, 6 figures, 1 table, 2 algorithms)

This paper contains 15 sections, 5 equations, 6 figures, 1 table, 2 algorithms.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Linear Temporal Logic
  4. Büchi Automata
  5. Agent Model
  6. Task Grammar - LTL$^\psi$
  7. Control Synthesis for LTL$^\psi$
  8. Approach
  9. Büchi Automaton for an LTL$^\psi$ Formula
  10. Agent Behavior for an LTL$^\psi$ Specification
  11. Finding Possible Individual Agent Bindings
  12. Agent Team Assignment
  13. Synthesis and Execution of Control and Synchronization Policies
  14. Results and Discussion
  15. Conclusion

Figures (6)

  • Figure 1: Agent partial model: (a) $\lambda_{\mathit{area}}$ (b) $\lambda_{\mathit{arm}}$ (c) $A_{green}$
  • Figure 2: Agriculture environment and initial agent states. The green, blue, and pink agents are stationary; the orientation of their sensors are indicated by the colored boxes.
  • Figure 3: $\mathcal{B}$ for $\varphi^\psi$ (Eq. \ref{['eq:task']}). The purple transitions illustrate a possible accepting trace.
  • Figure 4: A small portion of $\mathcal{G}_{green}$
  • Figure 5: The final step in the synchronized behavior of the agent team with their corresponding actions.
  • ...and 1 more figures