Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Cooperative Task Execution in Multi-Agent Systems

Karishma, Shrisha Rao

TL;DR

The paper tackles cooperative task execution in distributed multi-agent systems where tasks have dependencies and agents collaborate in groups. It introduces the Cooperative Execution Strategy (CES) that partitions agents into groups, assigns disjoint task subsets with corresponding rewards and dependencies, and supports both centralized and decentralized task distribution within groups. Theoretical contributions include a transitivity result for intra-group knowledge sharing and explicit waiting-time bounds $E[W] = \Theta(mkp^k)$ for maximum dependency degree $k$ and $E[W] = \Theta(mp^d)$ for fully connected graphs with $d=m-1$, illustrated alongside empirical comparisons between less-dependent and highly-dependent systems such as $G_{18}$ and $G_{40}$. The results show that many small groups outperform few large groups, with centralized control preferred for small task loads and decentralized control for large-scale tasks; task distribution favors independent sets in LDS and cross-group dependencies in HDS. Overall, the work provides actionable guidance on optimal group sizing and coordination strategies for scalable, dependency-aware multi-agent systems.

Abstract

We propose a multi-agent system that enables groups of agents to collaborate and work autonomously to execute tasks. Groups can work in a decentralized manner and can adapt to dynamic changes in the environment. Groups of agents solve assigned tasks by exploring the solution space cooperatively based on the highest reward first. The tasks have a dependency structure associated with them. We rigorously evaluated the performance of the system and the individual group performance using centralized and decentralized control approaches for task distribution. Based on the results, the centralized approach is more efficient for systems with a less-dependent system $G_{18}$ (a well-known program graph that contains $18$ nodes with few links), while the decentralized approach performs better for systems with a highly-dependent system $G_{40}$ (a program graph that contains $40$ highly interlinked nodes). We also evaluated task allocation to groups that do not have interdependence. Our findings reveal that there was significantly less difference in the number of tasks allocated to each group in a less-dependent system than in a highly-dependent one. The experimental results showed that a large number of small-size cooperative groups of agents unequivocally improved the system's performance compared to a small number of large-size cooperative groups of agents. Therefore, it is essential to identify the optimal group size for a system to enhance its performance.

Cooperative Task Execution in Multi-Agent Systems

TL;DR

The paper tackles cooperative task execution in distributed multi-agent systems where tasks have dependencies and agents collaborate in groups. It introduces the Cooperative Execution Strategy (CES) that partitions agents into groups, assigns disjoint task subsets with corresponding rewards and dependencies, and supports both centralized and decentralized task distribution within groups. Theoretical contributions include a transitivity result for intra-group knowledge sharing and explicit waiting-time bounds for maximum dependency degree and for fully connected graphs with , illustrated alongside empirical comparisons between less-dependent and highly-dependent systems such as and . The results show that many small groups outperform few large groups, with centralized control preferred for small task loads and decentralized control for large-scale tasks; task distribution favors independent sets in LDS and cross-group dependencies in HDS. Overall, the work provides actionable guidance on optimal group sizing and coordination strategies for scalable, dependency-aware multi-agent systems.

Abstract

We propose a multi-agent system that enables groups of agents to collaborate and work autonomously to execute tasks. Groups can work in a decentralized manner and can adapt to dynamic changes in the environment. Groups of agents solve assigned tasks by exploring the solution space cooperatively based on the highest reward first. The tasks have a dependency structure associated with them. We rigorously evaluated the performance of the system and the individual group performance using centralized and decentralized control approaches for task distribution. Based on the results, the centralized approach is more efficient for systems with a less-dependent system (a well-known program graph that contains nodes with few links), while the decentralized approach performs better for systems with a highly-dependent system (a program graph that contains highly interlinked nodes). We also evaluated task allocation to groups that do not have interdependence. Our findings reveal that there was significantly less difference in the number of tasks allocated to each group in a less-dependent system than in a highly-dependent one. The experimental results showed that a large number of small-size cooperative groups of agents unequivocally improved the system's performance compared to a small number of large-size cooperative groups of agents. Therefore, it is essential to identify the optimal group size for a system to enhance its performance.
Paper Structure (8 sections, 4 theorems, 8 equations, 3 figures, 4 tables, 1 algorithm)

This paper contains 8 sections, 4 theorems, 8 equations, 3 figures, 4 tables, 1 algorithm.

Table of Contents

  1. INTRODUCTION
  2. COOPERATIVE EXECUTION STRATEGY FOR GROUPS OF AGENTS
  3. System Model
  4. CES Algorithm
  5. RESULTS
  6. Mathematical Results
  7. Experimental Results
  8. CONCLUSIONS

Key Result

theorem 1.1

If $t_k \sim t_l$, agent $a_i$ knows the solution for task $t_k$:$(t_k, s_k) \in K(a_i)$ and two agents, $a_i$ and $a_j$ belong to the same group: ${a_i, a_j} \in g_p$, then $(t_l, s_k) \in K(a_j)$

Figures (3)

  • Figure 1: Task Dependency Graph $G_{10}$ for groups.
  • Figure 2: System performance for centralized and decentralized group approach.
  • Figure 3: Task distribution between groups of agents in LDS and HDS.

Theorems & Definitions (6)

  • theorem 1.1: Transitivity on Knowledge
  • theorem 1.3: Impact of Dependencies on Expected Waiting Time for LDS
  • proof
  • theorem 1.4: Impact of Dependencies on Expected Waiting Time for Fully Connected Graph
  • theorem 1.5: Optimal Group Size
  • proof