Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SwiftBot: A Decentralized Platform for LLM-Powered Federated Robotic Task Execution

YueMing Zhang, Shuai Xu, Zhengxiong Li, Fangtian Zhong, Xiaokun Yang, Hailu Xu

Abstract

Federated robotic task execution systems require bridging natural language instructions to distributed robot control while efficiently managing computational resources across heterogeneous edge devices without centralized coordination. Existing approaches face three limitations: rigid hand-coded planners requiring extensive domain engineering, centralized coordination that contradicts federated collaboration as robots scale, and static resource allocation failing to share containers across robots when workloads shift dynamically. We present SwiftBot, a federated task execution platform that integrates LLM-based task decomposition with intelligent container orchestration over a DHT overlay, enabling robots to collaboratively execute tasks without centralized control. SwiftBot achieves 94.3% decomposition accuracy across diverse tasks, reduces task startup latency by 1.5-5.4x and average training latency by 1.4-2.5x, and improves tail latency by 1.2-4.7x under high load through federated warm container migration. Evaluation on multimedia tasks validates that co-designing semantic understanding and federated resource management enables both flexibility and efficiency for robotic task control.

SwiftBot: A Decentralized Platform for LLM-Powered Federated Robotic Task Execution

Abstract

Federated robotic task execution systems require bridging natural language instructions to distributed robot control while efficiently managing computational resources across heterogeneous edge devices without centralized coordination. Existing approaches face three limitations: rigid hand-coded planners requiring extensive domain engineering, centralized coordination that contradicts federated collaboration as robots scale, and static resource allocation failing to share containers across robots when workloads shift dynamically. We present SwiftBot, a federated task execution platform that integrates LLM-based task decomposition with intelligent container orchestration over a DHT overlay, enabling robots to collaboratively execute tasks without centralized control. SwiftBot achieves 94.3% decomposition accuracy across diverse tasks, reduces task startup latency by 1.5-5.4x and average training latency by 1.4-2.5x, and improves tail latency by 1.2-4.7x under high load through federated warm container migration. Evaluation on multimedia tasks validates that co-designing semantic understanding and federated resource management enables both flexibility and efficiency for robotic task control.
Paper Structure (15 sections, 8 figures, 1 table, 1 algorithm)

This paper contains 15 sections, 8 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Motivation and Background
  3. From Single Robots to Federated Fleets
  4. Federated Learning for Collaborative Robotics
  5. System Design
  6. Task Decomposer
  7. Dynamic Task Manager
  8. Federated Warm-start Container Pool
  9. Evaluation
  10. LLM-Based Task Decomposition Performance
  11. Performance of Federated Applications
  12. Container Selection and Migration Performance
  13. Discussion
  14. Related Work
  15. Conclusion

Figures (8)

  • Figure 1: The overview of SwiftBot workflow and structure.
  • Figure 2: Reschedule a task based on runtime resources of robots and task needs.
  • Figure 3: Comparison of training accuracy under different client configurations in UCF101 dataset.
  • Figure 4: Comparison of training accuracy under different client configurations in LibriSpeech dataset.
  • Figure 5: Comparison of training loss in FedAvg vs. SwiftBot.
  • ...and 3 more figures